Rodney Brooks

Robots, AI, and other stuff

My Dated Predictions

With all new technologies there are predictions of how good it will be for humankind, or how bad it will be. A common thread that I have observed is how people tend to underestimate how long new technologies will take to be adopted after proof of concept demonstrations. I pointed to this as the seventh of seven deadly sins of predicting the future of AI.

For example, recently the early techno-utopianism of the Internet providing a voice to everyone and thus blocking the ability of individuals to be controlled by governments has turned to depression about how it just did not work out that way. And there has been discussion of how the good future we thought we were promised is taking much longer to be deployed than we had ever imagined. This is precisely a realization of the early optimism about how things would be deployed and used did just not turn out to be.

Over the last few months I have been throwing a little cold water over what I consider to be current hype around Artificial Intelligence (AI) and Machine Learning (ML). However, I do not think that I am a techno-pessimist. Rather, I think of myself as a techno-realist.

In my view having ideas is easy. Turning them into reality is hard. Turning them into being deployed at scale is even harder. And in evaluating the likelihood of success at that I think it is possible to sort technology and technology deployment ideas into a spectrum running from relatively easier to very hard.

But simply spouting off about this is rather easy to do as there is no responsibility for being right or wrong. That applies not just to me, but to pundits ranging from physicists to entrepreneurs to academics, who are making wild predictions about AI and ML.

It is the New Year  and there will be many predictions about what will happen in the coming year. I am going to take this opportunity to make predictions myself, not just about the coming year, but rather the next thirty two years. I am going to write them in this blog with explicit dates attached to them. Hence they are my dated predictions. And they will be here on this blog and copies that live on elsewhere in cyberspace for all to see. I am going to take responsibility for what I say, and make it so that people can hold me to whether I turn out to be right or wrong. If I am unfortunate, some of my predictions will at some point seem rather dated!

I chose thirty two years as I will then be 95 years old, and I suspect I’ll be a little too exhausted by then to carry on arguments about why I was right or wrong on particular points. And 32 is a power of 2, so that’s always a good thing. So the furtherest out date I am going to consider is January 1st, 2050. And that also means that I am only predicting things for exactly the first half of this century (or at least for the first half of the years starting with “20” — there is a whole argument to be had here into which I am not going to get).

I specify dates in three different ways:

NIML meaning “Not In My Lifetime, i.e., not until after January 1st, 2050

NET some date, meaning “No Earlier Than” that date.

BY some date, meaning “By” that date.

Sometimes I will give both a NET and a BY for a single prediction, establishing a window in which I believe it will happen.


I am going to try to be very precise about what I am predicting and when. Now in reality precision on defining what I am predicting is almost impossible. Nevertheless I will try.

I had an experience very recently that made me realize just how hard people will try, when challenged, to hold their preconceived notions about technologies and the cornucopia they will provide to humanity. I tweeted out the following (@rodneyabrooks):

When humans next land on the Moon it will be with the help of many, many, Artificial Intelligence and Machine Learning systems.

Last time we got to the Moon and back without AI or ML.

My intent with this tweet was to say that although AI and ML are today very powerful and useful, it does not mean that they are the only way to do things, and it is worth remembering that. They don’t necessarily mean that suddenly everything has changed in the world in some magical way1.

One of the responses to this tweet, which itself was retweeted many, many times, was that Kalman filtering was used to track the spacecraft (completely true), that Kalman filtering uses Bayesian updating (completely true), and that therefore Kalman filtering is an instance of machine learning (complete non sequitur) and that therefore machine learning was used to get to the Moon (a valid inference based on a non-sequitur, and completely wrong).  When anyone says Machine Learning these days (and indeed since the introduction of the term in 1959 by Arthur Samuel (see my post on ML for details)) they mean using examples in some way to induce a representation of some concept that can later be used to select a label or action, based on an input and that saved learned material. Kalman filtering uses multiple data points from a particular process to get a good estimate of what the data is really saying. It does not save anything for later to be used for a similar problem at some future time. So, no, it is not Machine Learning, and no, we did not use Machine Learning to get to the Moon last time, no matter how much you want to believe that Machine Learning is the key to all technological progress.

That is why I am going to try to be very specific about what I mean by my predictions, and why, no doubt, I will need to argue back to many people who will want to claim that the things I predict will not happen before some future time have already happened. I predict that people will be making such claims!

What is Easy and What is Hard?

Building electric cars and reusable rockets is easy. Building flying cars, or a hyperloop system (or a palletized underground car transport network underground) is hard.

What makes the difference?

Cars have been around, and mass produced, for well over a century. If you want to build electric cars rather than gasoline cars, you do not have to invent too much stuff, and figure out how to deploy it at scale.

There has been over a hundred years of engineering and production of windscreen wipers, brakes, wheels, tires, steering systems, windows that can go up and down, car seats, a chassis, and much more. There have even been well over 20 years of large scale production of digitalized drive trains.

To build electric cars at scale, and at a competitive price, and with good range, you may have to be very clever, and well capitalized. But there is an awful lot of the car that you do not need to change. For the majority of the car there are plenty of people around who have worked on those components for decades, and plenty of manufacturing expertise for building the components and assembly.

Although reusable rockets sounds revolutionary there is again prior art and experience. All liquid fuel rockets today owe their major components and capabilities to the V-2 rockets of Wernher von Braun, built for Hitler. It was liquid fueled with high flow turbopumps (580 horsepower!), it used the fuel to cool parts of the engine, and it carried its own liquid oxygen so that it could fly above the atmosphere. It first did so just over 75 years ago. And it was mass produced, with 5,200 of them being built, using slave labor, in just two years.

Since then over 20 different liquid fueled rocket families have been developed around the world, some with over 50 years of operational use, and hundreds of different configurations within those families. Many variations in parameters and trade offs have been examined. Soyuz rockets, a fifty year old family, all lift off with twenty liquid fueled thrust chambers burning. In the Delta family, the Delta IV configuration has a “Heavy” variant, three essentially identical cores in a horizontal line, where the cores are all a first stage of the earlier single core Delta IV.

The technology for soft landing on Earth using jet engine thrusters has been around since 1950s with the Rolls Royce “flying bedstead”, with the later, at large scale, Harrier fighter jet taking off and landing vertically. A rocket engine for vertical landing was used, without atmosphere, for the manned lunar landings on the Moon, starting in 1969.

Today’s Falcon rocket uses grid fins to  steer the first stage when it is returning to the launch site or recovery barge to soft land. These were first developed theoretically in Russia in the 1950’s by Sergey Belotserkovskiy and have been used since the 1970’s for many missiles, both ballistic and others, guided bombs, cruise missiles, and for the emergency escape system for manned Soyuz capsules.

There has been a lot of money spent on developing rockets and this has lead to many useable technologies, lots of know how, and lots of flight experience.

None of this is to say that developing at scale electric cars or reusable rockets is not brave, hard, and incredibly inventive work. It does however build on large bodies of prior work, and therefore it is more likely to succeed. There is experience out there. There are known solutions to many, many, but not all, problems that will arise. Seemingly revolutionary concepts can arise from clusters of hard and brilliantly thought out evolutionary ideas, along with the braveness and determination to undertake them.

We can make estimates about these technologies being technically successful and deployable at scale with some confidence.

For completely new ideas, however, it is much harder to predict with confidence that the technologies will become deployable in any particular amount of time.

There have been sustained projects working on problems of practical nuclear fusion reactors for power generation since the 1950’s. We know that sustained nuclear fusion “works”. That is how our Sun and every other star shines. And humans first produced short time scale nuclear fusion with the first full scale thermonuclear bomb, “Ivy Mike”, being detonated 65 years ago. But we have not yet figured out how to make nuclear fusion practical for anything besides bombs, and I do not think many people would believe any predicted date for at scale practical fusion power generation. It is a really hard problem.

The hyperloop concept has attracted a bunch of start ups and capital for them, though there has been nothing close in concept that has ever been demonstrated, let alone operated at scale. So besides figuring out how to develop ultrastable cylinders that go for hundreds of miles, containing capsules that are accelerated by external air pressure traveling at hundreds of miles per hour while containing living meat of the human variety there are many, many mundane things to be developed.

One of the many challenges is how to seal the capsules and provide entirely self contained life support within for the duration of the journey. Also the capsules must be able to go past stations at which they are not stopping in a stable manner, so stations will need to be optionally sealed off from the tube for a through capsule, while allowing physical ingress and egress for passengers whose capsule has stopped at the station. There will need to be procedures for when a capsule gets stuck a hundred miles from the nearest station. There will need to be communications with the capsule, even though it is in a pretty good Faraday cage. There will need to be the right seats and restraints developed for the safety of the passengers. There will need to user experience elements developed for the sanity of the passengers while they are being whizzed at ultra high speed in windowless capsules. And then there are route rights, earthquake protection, dealing with containing cylinder distortions just because of the centimeter or so of drift induced along the route in the course of year just due to normal smooth deformations of our tectonic plates. And then there are pricing models, and getting insurance, and figuring out how that interacts with individual passenger insurance. Etc., etc.

There will need to be many, many new technologies and new designs developed for every aspect of the hyperloop. None of them will have existed before. None of them have been demonstrated, nor even enumerated as of today. It is going to take a long time to figure all these things out and build a stable system around them, and to do all the engineering needed on all the components. And it is going to be a hard psychological sell for passengers to ride in these windowless high speed systems, so even when all the technology challenges have been knocked down there will still be the challenge of pace of adoption.

So…while there might be some demonstration of some significance in the next 32 years I am confident in saying that there will be no commercial viable passenger carrying systems for hyperloop within that time frame.

I use this framework in trying to predict timing on various technological innovations. If something has not even been demonstrated yet in the lab, even though the physics says that it will be good to go, then I think it is a long, long way off. If it has been demonstrated in prototypes only, then it is still a long way off. If there are versions of it deployed at scale already, and most of what needs doing is evolutionary, then it may happen before too long. But then again, no-one may want to adopt it, so that will slow things down no matter how much enthusiasm there is by the technologists involved in developing it.


Adoption of new things in technology takes much longer than one might expect. The original version of the Internet used 32 bit addressing, allowing only 4 billion unique address for all devices on the network, and using a protocol called IPv4, Internet Protocol version 4.  But by the early 1990’s it was recognized that with all the devices that would soon join the network (not just personal devices but so many other things like electricity meters, industrial sensors, traffic sensor and control, TVs, light switches(!), etc., etc.) the world would soon run out of address space.

By 1996 a new protocol,  IPv6, Internet Protocol version 6, had been defined, increasing the address space to 128 bits from 32 bits, allowing for 7.9\times 10^{28} more devices on the network.

Since 1996 there have been various goal dates specified for when all network traffic should use IPv6 rather than IPv4. In 2010 the target date was 2012. In 2014 fully 99% of all network traffic was still using IPv4 with many, many clever edge systems to cram much more than 4 billion devices into a 4 billion device address space. By the end of 2017 various categories of network traffic running on IPv6 ranged from under 2% to just over 20%. It is still a long way from full adoption of IPv6.

There were no technical things stopping the adoption of IPv6, in fact quite the opposite. As the number of devices that wanted to connect to the Internet grew there had to be many very clever innovations and work arounds in order to limp along with IPv4 rather than adopt IPv6.

Using my heuristics (rate of replacement of equipment, maturity of technical solutions, real need for what it provides, etc.) that I use to make my predictions in this post, I would have thought that IPv6 would have been universal by 2010 or so. I would have been wildly over optimistic about it.


SpaceX first announced their Falcon Heavy rocket in April 2011, broke ground on their Vandenberg AFB, California, launch pad for it in June 2011, and expected a maiden flight in 2013. The rocket was first moved to a launch pad on December 28, 2017, at pad 39A at the Kennedy Space Center in Florida. It is now expected to fly in 2018. Development time has stretched from two years to seven years. So far.

It always takes longer than you think. It just does.


The first three entries in the table below are about flying cars. I am pretty sure that practical flying cars will need to be largely self driving while flying, so they sort of fit the category. By flying cars I mean vehicles that can be driven anywhere a car can be driven. Otherwise it is not a car. And I mean that a person who does not have a pilots license, but does perhaps have a few hours of special training, can get into wearing normal clothing that would be appropriate to wear at an office, and is able to travel 100 miles, say, with much of the journey in the air. It should require no previous arrangement for the journey, no special filing of plans, nothing beyond using a maps like app on a smartphone in order to know the route to get to the destination. In other words, apart from a little extra training it should be just like an average person today using a conventional automobile to travel 100 miles.

Now let’s talk about self driving cars, or driverless cars. I wrote two blog posts early in 2017 about driverless cars. The first talked about unexpected consequences of driverless cars, in that pedestrians and other drivers will interact with them in different ways than they do with cars with drivers in them, and how the cars may bring out anti-social behavior in humans outside of them. It also pointed out  that owners of individual driverless cars may use them in new ways that they could never use a regular car, sometime succumbing to anti-social behavior themselves. The second post was about edge cases in urban environments where there are temporary signs that drivers must read, where on a regular basis it is impossible to drive according to the letter of the law, where mobility as a service will need to figure out how much control a passenger is allowed to have, and where police and tow truck drivers must interact with these cars, and the normal human to human interaction with drivers will no longer be present nor subjugatable by a position of authority.

For me it seems clear that driverless cars are not going to simply be the same sorts of cars as normal cars, but simply without human drivers. They are going to be fundamentally different beasts with different use modes, and different ways of fitting into the world.

Horseless carriages did not simply one for one replace horse drawn carriages. Instead they demanded a whole new infrastructure of paved roads, a completely new ownership model, a different utilization model, completely different fueling and maintenance procedures, a different rate of death for occupants, a different level of convenience, and ultimately they lead to a very different structure for cities as they enabled suburbia.

I think the popular interpretation is that driverless cars will simply replace cars with human drivers, one for one. I do not think that is going to happen at all. Instead our cities will be changed with special lanes for driverless cars, geo-fencing of where they can be and where cars driven by humans can be, a change in the norm for pick up and drop off location flexibility, changes to parking regulations, and in general all sorts of small incremental modifications to our cities.

But first let’s talk about the rate of adoption of driverless cars.

As I pointed out in my seven deadly sins post, in 1987 Ernst Dickmanns and his team at the Bundeswehr University in Munich had their autonomous van drive at 90 kilometers per hour (56mph) for 20 kilometers (12 miles) on a public freeway. Of course there were people inside the van but they had their hands off the controls. For the last 30 years researchers have been improving the ability of cars to drive on public roads, but it has mostly been about the driving, with very little about the interaction, the pick up and drop off of people, the interface with other services and restrictions, and with non-driving passengers inside the cars. All of these will be important.

From one point of view it has been slow, slow, slow incremental progress over the last thirty years, even though the work has been focused on only a small part of the problem. Just about a year ago I saw a tweet which I loved, which said something like “The customers knew that they had gotten a driverless Uber as there were two people in the front seat instead of just one.”. It is only just in the last few weeks that have started seeing actual unoccupied cars on public roads, from Waymo in Phoenix, Arizona. A tweet about this story referred to them as being the first “driverless driverless cars”…

But adoption is still a ways off. The price of the sensors still needs to come way down, and all the operational things about how the cars will be used and interface with passengers still needs to be worked out, let alone all the actual regulatory and liability environment under which they will operate needs to be put in place. Within some constraints, all these things will eventually be solved. But it is going to be much slower than many expect.

The true test of the viability of driverless cars will be when they are not just in testing or in demonstration, but when the owners of driverless taxis or ride sharing services or parking garages for end consumer self driving cars are actually making money at it. This will happen only gradually and in restricted geographies and markets to start with. My milestone predictions below are not about demonstrations, but about viable sustainable businesses. Without them the deployment of driverless cars will never really take off.

I think the under discussed reality of how driverless cars will get adopted is through geo fencing of where certain activities of those cars can take place, without any human driven cars in that vicinity. Furthermore applications of driverless cars will initially be restricted to certain cities and even areas within those cities, and perhaps even certain times of day and in certain weather conditions. It may be that for quite a while the cars for the first mobility as a service driverless cars (e.g., for Uber and Lyft like services) will only operate in a driverless mode some of the time, and at other times will need to have hired human drivers.

[Self Driving Cars]
A flying car can be purchased by any US resident if they have enough money.NET 2036There is a real possibility that this will not happen at all by 2050.
Flying cars reach 0.01% of US total cars.NET 2042That would be about 26,000 flying cars given today's total.
Flying cars reach 0.1% of US total cars.NIML
First dedicated lane where only cars in truly driverless mode are allowed on a public freeway.
NET 2021This is a bit like current day HOV lanes. My bet is the left most lane on 101 between SF and Silicon Valley (currently largely the domain of speeding Teslas in any case). People will have to have their hands on the wheel until the car is in the dedicated lane.
Such a dedicated lane where the cars communicate and drive with reduced spacing at higher speed than people are allowed to driveNET 2024
First driverless "taxi" service in a major US city, with dedicated pick up and drop off points, and restrictions on weather and time of day.NET 2022The pick up and drop off points will not be parking spots, but like bus stops they will be marked and restricted for that purpose only.
Such "taxi" services where the cars are also used with drivers at other times and with extended geography, in 10 major US citiesNET 2025A key predictor here is when the sensors get cheap enough that using the car with a driver and not using those sensors still makes economic sense.
Such "taxi" service as above in 50 of the 100 biggest US cities.NET 2028It will be a very slow start and roll out. The designated pick up and drop off points may be used by multiple vendors, with communication between them in order to schedule cars in and out.
Dedicated driverless package delivery vehicles in very restricted geographies of a major US city.NET 2023The geographies will have to be where the roads are wide enough for other drivers to get around stopped vehicles.
A (profitable) parking garage where certain brands of cars can be left and picked up at the entrance and they will go park themselves in a human free environment.NET 2023The economic incentive is much higher parking density, and it will require communication between the cars and the garage infrastructure.
A driverless "taxi" service in a major US city with arbitrary pick and drop off locations, even in a restricted geographical area.
NET 2032This is what Uber, Lyft, and conventional taxi services can do today.
Driverless taxi services operating on all streets in Cambridgeport, MA, on Greenwich Village, NY, NET 2035Unless parking and human drivers are banned from those areas before then.
A major city bans parking and cars with drivers from a non-trivial portion of a city so that driverless cars have free reign in that area.NET 2027
BY 2031
This will be the starting point for a turning of the tide towards driverless cars.
The majority of US cities have the majority of their downtown under such rules.NET 2045
Electric cars hit 30% of US car sales.NET 2027
Electric car sales in the US make up essentially 100% of the sales.NET 2038
Individually owned cars can go underground onto a pallet and be whisked underground to another location in a city at more than 100mph.NIMLThere might be some small demonstration projects, but they will be just that, not real, viable mass market services.
First time that a car equipped with some version of a solution for the trolley problem is involved in an accident where it is practically invoked.NIMLRecall that a variation of this was a key plot aspect in the movie "I, Robot", where a robot had rescued the Will Smith character after a car accident at the expense of letting a young girl die.
Predictions about ROBOTICS, AI and ML

Those of you who have been reading my series of blog posts on the future of Robotics and Artificial Intelligence know that I am more sanguine about how fast things will deploy at scale in the real world than many cheerleaders and fear mongers might believe. My predictions here are tempered by that sanguinity.

Some of these predictions are about the public perception of AI (that has been the single biggest thing that has changed in the field in the last three years), some are about technical ideas, and some are about deployments.

[AI and ML]
Academic rumblings about the limits of Deep LearningBY 2017Oh, this is already happening... the pace will pick up.
The technical press starts reporting about limits of Deep Learning, and limits of reinforcement learning of game play.BY 2018
The popular press starts having stories that the era of Deep Learning is over.BY 2020
VCs figure out that for an investment to pay off there needs to be something more than "X + Deep Learning".NET 2021I am being a little cynical here, and of course there will be no way to know when things change exactly.
Emergence of the generally agreed upon "next big thing" in AI beyond deep learning.NET 2023
BY 2027
Whatever this turns out to be, it will be something that someone is already working on, and there are already published papers about it. There will be many claims on this title earlier than 2023, but none of them will pan out.
The press, and researchers, generally mature beyond the so-called "Turing Test" and Asimov's three laws as valid measures of progress in AI and ML.NET 2022I wish, I really wish.
Dexterous robot hands generally available.NET 2030
BY 2040 (I hope!)
Despite some impressive lab demonstrations we have not actually seen any improvement in widely deployed robotic hands or end effectors in the last 40 years.
A robot that can navigate around just about any US home, with its steps, its clutter, its narrow pathways between furniture, etc.Lab demo: NET 2026
Expensive product: NET 2030
Affordable product: NET 2035
What is easy for humans is still very, very hard for robots.
A robot that can provide physical assistance to the elderly over multiple tasks (e.g., getting into and out of bed, washing, using the toilet, etc.) rather than just a point solution.NET 2028There may be point solution robots before that. But soon the houses of the elderly will be cluttered with too many robots.
A robot that can carry out the last 10 yards of delivery, getting from a vehicle into a house and putting the package inside the front door.Lab demo: NET 2025
Deployed systems: NET 2028
A conversational agent that both carries long term context, and does not easily fall into recognizable and repeated patterns.Lab demo: NET 2023
Deployed systems: 2025
Deployment platforms already exist (e.g., Google Home and Amazon Echo) so it will be a fast track from lab demo to wide spread deployment.
An AI system with an ongoing existence (no day is the repeat of another day as it currently is for all AI systems) at the level of a mouse.NET 2030I will need a whole new blog post to explain this...
A robot that seems as intelligent, as attentive, and as faithful, as a dog.NET 2048This is so much harder than most people imagine it to be--many think we are already there; I say we are not at all there.
A robot that has any real idea about its own existence, or the existence of humans in the way that a six year old understands humans.NIML

These predictions may seem a little random and disjointed. And they are. But that is the way progress is going to be made in Robotics, AI, and ML. There is not going to be a general intelligence that can suddenly do all sorts of things that humans (or chimpanzees) can do. It is going to be point solutions for a long, long time to come.

Building human level intelligence and human level physical capability is really, really hard. There has been a little tiny burst of progress over the last five years, and too many people think it is all done. In reality we are less than 1% of the way there, with no real intellectual ideas yet on how to get to 5%. And yes, I made up those percentages and can not really justify them. I may well have inflated them by a factor of 10 or more, and for that I apologize.


I have been a fan of spaceflight since my childhood, when every week my father would fly from Adelaide to Woomera, South Australia, to work on the first stage engines of a European satellite launch initiative know as Europa. Every couple of months I would go with him on a Friday evening to meetings of a club of enthusiasts where they would have the latest film footage from NASA which would be projected and discussed.

I decided back then that my life goal was to eventually live on another planet. So far my major progress towards that goal is to have not died on Earth before leaving. In my realistic moments I realize now that I may eventually fail at my goal.

So here are my predictions about space travel. Not as optimistic as wish I could be. But, realistic, I think.

Next launch of people (test pilots/engineers) on a sub-orbital flight by a private company.BY 2018
A few handfuls of customers, paying for those flights.NET 2020
A regular sub weekly cadence of such flights.NET 2022
BY 2026
Regular paying customer orbital flights.NET 2027Russia offered paid flights to the ISS, but there were only 8 such flights (7 different tourists). They are now suspended indefinitely.
Next launch of people into orbit on a US booster.NET 2019
BY 2021
BY 2022 (2 different companies)
Current schedule says 2018.
Two paying customers go on a loop around the Moon, launch on Falcon Heavy.NET 2020The most recent prediction has been 4th quarter 2018. That is not going to happen.
Land cargo on Mars for humans to use at a later date
NET 2026SpaceX has said by 2022. I think 2026 is optimistic but it might be pushed to happen as a statement that it can be done, rather than for an pressing practical reason.
Humans on Mars make use of cargo previously landed there.NET 2032Sorry, it is just going to take longer than every one expects.
First "permanent" human colony on Mars.NET 2036It will be magical for the human race if this happens by then. It will truly inspire us all.
Point to point transport on Earth in an hour or so (using a BF rocket).NIMLThis will not happen without some major new breakthrough of which we currently have no inkling.
Regular service of Hyperloop between two cities.NIMLI can't help but be reminded of when Chuck Yeager described the Mercury program as "Spam in a can".

1AI and ML have been around for a long time already. I have been in pursuit of their magic for a long time. I have worked in both Artificial Intelligence and Machine Learning for over forty years. My 1977 Master’s thesis used Markov chains to prove the convergence of a particular machine learning algorithm. It was an abysmally terrible thesis.

90 comments on “My Dated Predictions”

      1. This qualifies for me as a “flying car”. But since it has to take off and land at an airport I can’t quite see the utility of it being worth the engineering penalties to make it both a real car and a real airplane.

      2. The engineering trade-offs are massive, and that may be what dooms it. Put simply, if light enough to fly well, it is hard to make it strong enough to meet road-safety crashworthiness standards.

        But it does not need an airport to take off: any football field or soccer pitch will do. In the spirit of full disclosure, I am an advisor to the firm, and organizations like Medecins san Frontieres are interested, because it can fly almost anywhere (to deliver vaccines e.g.) without requiring a specialized crew, and is able to be driven somewhere were it can be kept safe from unsavory elements (e.g. a regular garage), as opposed to having to be guarded somewhere (like an airport).

  1. A couple of weeks ago I predicted ( the following about the adoption of EVs.


    My guess is that once EVs reach 25-30% of the market — and become simply another option — the transition to 100% will happen quite quickly. After all, they are cheaper, more fun to drive, etc. At the current rate, that may still take another 15-20 years, which puts as at, say 2035.

    But my (perhaps optimistic) bet is that it will happen faster than that and that EVs will be 50% of the US new car market by 2030 — and perhaps sooner elsewhere, like China.

    1. I believe it’s important to differentiate between “average penetration rates” and “urban penetration rates”, where the latter will define the tipping point, as well as drive the societal changes. If one speculates that the vast majority of early adopters of EV will be found in urban and suburban areas (because that’s where the wealth, eco-consciousness, and regulations are largely concentrated), then the relative penetration can be 2.3x higher (based on W. European analysis) in these areas compared to the “average”. So if you consider that 30% of new car sales would be EV, then this actually would indicate a e.g. 70% rate in urban areas.

      There are many reasons this is important, but as EV penetration rates in urban areas increase, fuel sites will start to fail as EV owners start to charge their cars at home, at the office, on the street, at a shopping mall, etc. – which means lower “energy sales” for a fuel site as well as fewer visits to drive the high margin convenience store sales. Once fuel sites start to fail, it increases the desirability of EVs, as it becomes increasingly difficult to fuel your ICE (internal combustion engine) auto, and the price of fuel may actually increase in urban areas (due to the drop in scale effects) even in an environment of lower oil prices.

      This acceleration towards the tipping point in urban areas (which will also be supported by regulations in the larger cities), will then drive an increased rate of adoption for the rest of a country. In my analysis based on a broad swath of research from many different sources (including oil majors), I came to the conclusion that we’ll start seeing average urban fuel site revenues (and profits) drop by roughly 30% by 2025… which is a decline that can hardly be swallowed by the vast majority of fuel sites.

      So in this case, the secondary impacts of EV adoption will lead to a further acceleration of EV penetration. And when autonomous cars do make their presence felt in the urban environment, then this effect is compounded.

  2. I agree with almost everything you write here *EXCEPT* I that would substitute every instance of “deep learning” by “supervised learning”.
    The general idea if deep learning (assembling parameterized functional blocks and optimizing them through gradient-based methods) is simply not going away.
    One could have argued that the steam engine would go away in the early 20th century. They didn’t quite go away, but even if the did, the whole idea of a thermal engine didn’t. Deep learning is like thermal engines, not like steam engines.

    1. Thanks Yann, I did not mean to suggest that either Deep Learning or the general idea will go away. Rather I was saying that the hype bubble around them will burst. All the new techniques of the last five years are here to stay, and will continue to be useful. BUT, there will be a new poster child for the “power of AI” and a new hype bubble around something different in a few years. And the applications for graduate degrees at the top schools will mention this new poster child with higher frequency than “Deep Learning” (the current top phrase in graduate applications at all the big AI/CS schools).

  3. Rodney, another fantastic article, please keep them coming.

    I bought my family a Google Home for Christmas and although it was received well, its limitations were quickly reached. e.g. trying sending an email.

    This prompted my six year old daughter to ask “Daddy is it true that robots will do all the work, when I’m grown up?”

    I said “I don’t know, but there’s a clever man called Rodney, I’ll ask him.”

    1. I’m guessing no reply means my question is too speculative, but then again isn’t this entire blog post speculation?

      “We can refuse to understand the mechanics behind a theory and instead accept the word of an authority figure. If we fail to do the math on our own, we lose agency and the ability to develop an even more nuanced understanding of how the world works.” – Seth Godin

      Has anyone done this course?–ud730
      Seems Deep Learning is the shiney new buzzword these days.

      1. Rodney,

        My apologies, I will try to be more clear in the future and stick to one train of thought per comment.

        OK I was making three points, albeit badly. I will now try to redress my previous comments, so here goes:

        Post 1.) I read your fascinating article, and coincidentally the same day my 6 year old daughter asked me the question “Will robots do all the work when I’m grown up?”

        Based upon your prediction of:

        “A robot that can carry out the last 10 yards of delivery, getting from a vehicle into a house and putting the package inside the front door. – NET 2028.”

        Then my daughter would be approximately 16 years old (grown up for sure!) when a robot could do all “the work”.

        Three days passed without you replying. That’s not a criticism just a statement. I do appreciate you’re busy and this blog is “for fun” but I couldn’t help but wonder why you’d managed to answer other people ‘s comments in those three days, but not mine? My first post was clear enough, no???

        This prompted me to make a second follow-up comment, which I fully admit is fragmented, my apologies for that.

        Post 2.) I was always taught to never take another person’s opinion as fact, no matter how much of an “expert” they might be. That instead one should endeavour to corroborate facts independently for oneself. Hence the Seth Godin quote.

        “We can refuse to understand the mechanics behind a theory and instead accept the word of an authority figure. If we fail to do the math on our own, we lose agency and the ability to develop an even more nuanced understanding of how the world works.” – Seth Godin

        In seemed relevant Why? Well as you’ve pointed out numerous times before there’s a lot of hype surrounding AI/ML in the media (an authority figure) at the moment. I was alluding (badly) that we should all try to find credible sources of information and further our own understanding.

        Credible sources for me, are sites such as your blog and Sebastian Thrun’s Udacity.

        Other people reading this blog are likely to be interested in AI/ML, so I thought it would be a good place to ask if the Udacity course was any good. I sure it is, I was just canvassing opinions.

        I hope this follow up comment is taken in the good nature it is intended, my apologies for having to make you read it.

        I look forward to your next blog post with earnest anticipation.

        Kind Regards,


      2. I haven’t looked at the Udacity course that you mention, but in general their courses have solid reputations. Following your quote from Seth Goldin you will have to take it yourself and evaluate it!

  4. Very sensible. Re conversational systems, it’s timescale of interaction that tells the tale. And scoping: many efforts are focused at local chit-chat rather than conversations, the latter which include establishing and monitoring mutual goals. So they look okay for a short time, but aren’t coherent over longer timescales.

  5. Thanks for very interesting post. I think your Mars first landing date is a little optimistic and your emergence of intelligent AI a little pessimistic but there are so many variables that having any degree of accuracy is not possible.

    I was always hoping that I might live to meet an android that was self aware but it doesn’t seem possible now. I’ll make a guess at 2140 for the first ‘human’ android.

    I agree with specially designated lanes for self driving cars. The scenario I put forward was off two cars meeting head on on a wet dark evening on a narrow English country lane.

    With a human driver you just reverse back to the nearest passing passing point or judge if by going up on the verge you can squeeze past. I can’t see any AI being able to do that for a considerable amount of time especially if the passing point is a bit overgrown and muddy due to all the rain.

    If my parents and grandparents are anything to go by I might live for another 30 years or perhaps a little longer – but although the World will be different in 2048 – it won’t be that different.

    Looking back to 1988 there are many recognizable things – mobile phones – home PCs – Golfs and BMW’s will never change outwardly that much and neither will jets. Look at a 737 or 757 from 1988 – they are still the same as today although they would have had a glass cockpit made up of CRT screens.

    I’m hoping that quantum encryption will have made everything safe and in say 20 years time you might be able to tap into the power of a centrally based quantum computer for all sorts of applications.

  6. Bravo on your latest essay! Heartily agree and love your use of NIML!

    In realized, last quarter, that the current “state of the art” in the use of “AI” in business is today about as mature as Computer Science was in 1958.

    Helping my clients understand what this means for them and their plans will be my primary focus in 2018.

  7. I like your predictions about self driving vehicules and how a global use of such technologies involves a major change of infrastructures. However, your pragmatism sounds a little pessimistic. 32 years is also almost the interval between the commercialization of the first analog cell phones or the apple IIC and the release of the Iphone X. We have seen a major change of communication infrastructures and technolgies (internet/cells phones/GPS) that have deeply modify our way to socialize and use media. Don’t you think another large paradigm change could not happened in the next 32 years ?

    1. Yes, there probably will be vary large changes, and exactly what they look like are hard to predict. But I am more pessimistic about the time line than you are as I think there is a different decision making process than in your analogy of the changes from analog cell phones to iPhone X. In that case it could be done along side the existing infrastructure so you didn’t have to ask people to give up what they already had in order to provide something new. They co-existed. In the case of cities it will require re-apportioning a finite amount of geography, so something will have to shut down or get squeezed in order to add something new. And each decision will be different and will need to be made in each city, and in every city lots of people will be upset that they are losing something so that the “elite” (and it will be the elite initially) get something new. Ugly. It is going to take a lot of time I believe.

  8. Very interesting… Having been in computers, and the fringes of AI and ML since the late 70’s on, the dates seem reasonable, even a bit optimistic!

  9. Great article. Really enjoyed reading it. Following on the last comments, agree that replacing current manned transport forms with evolved autonomous ones in a safe manner will undoubtedly take much longer than what is being conveid to mass population.
    I would be much more realized if I was able to see in my lifetime any breakthroughs in AI and ML to help out fields such as particle physics that would allows us to teletransport objects between two locations. That would be something.

  10. Agree on most of this but I think you’ve got the order of dog-level intelligence and dextrous hands the wrong way round. In evolution, dog-level intelligence preceded dextrous manipulation by many millions of years and requires a lot less neural machinery. Human-level hand-eye co-ordination and in-hand manipulation capabilities are unique to our species and a large part of the reason for our relatively large brains. Perhaps you’re not specifying human-level dexterity for hands, but even to match primate abilities is going to be difficult.
    The hardest part of dog-level intelligence will be dog-level dexterity, dog-level social intelligence (attentiveness and fidelity) is not too hard, particularly given the “seems as” qualification which allows developers to exploit short-cuts. I think people could have a pet-like social experience with robots in the next five years. Which part of this problem is so hard that we are thirty years away?

    1. Hi Tony,
      I debated with myself putting in something about dog level dexterity. It is not something people usually talk about but it can be quite amazing. There is dexterity just with their paws, and then additionally with their mouth, and one assume making some use of their tongues, besides their jaws and teeth.

      I used NET dates for both dogs and dexterous hands, so in some sense did not specify an actual order. It could be that in the technological world there is a solution to dexterity which exceeds that of a dog’s dexterity without solving all the other open dogness capabilities.

      I think we will see more resources put into going after robot dexterity than we will see in going after dogness.

      1. Hi would you like to be licked by a robot dog? That in itself could present some interesting challenges.

      2. Dogs are remarkable at how they are able to manipulate things with their mouths, and one must assume with their tongue as part of that.

  11. A very thoughtful article Rodney. A very similar list could be made, unfortunately, for brain-machine interfaces (BMI) to enhance cognition or memory for any and all of us.

    Given the distance between neurons and the scalp, and the linear superposition of the EM field generated by these neurons and associated cells, accessing the cellular level from outside the brain is enormously challenging, greatly limiting any BMI to bulk-tissue signals and crude operations.

    This leaves invasive approaches which require trepanation, via drilling burr holes through the skull and the intervening meninges, or craniotomy. Given the attendant risks, no neurosurgeon in the Developed World will undertake such an operation in healthy individuals. Thus, the large-scale usage of such devices in the general population belongs into your dreaded NIML category (I’m about as old as you are).

    There is a lot of exciting proof-of-concept progress in enabling quadriplegics to regain some functionality but, so far, none of these BMI devices function under real-world conditions (e.g., on a daily basis, in the home of the patients, to improve their quality of life).


    1. Hi Christof!

      Yes, somethings turn out to be harder than we had hoped, or at least to look like they will take longer than we now see we might have left.

      But in the meantime we keep trying!

      All best to you,

  12. Kudos to you for your pragmatic predictions. It’s very difficult and lonely to be the realist/contrarian when so many others are swept up in the hype. I’ve always wondered about the effects of the hype cycle. Is the initial hype necessary to drum up interest and then investment? What if the expectations for payoff were more realistic from the beginning – both in duration and total investment required? Would we have taken the same approach? In other words, does the hype cycle represent to most efficient way to make something happen, or is there a better way?

  13. Hi Rodney

    I understand that my comparison with analog cell phones was not so good as internet and communication technologies has grown smoothly beside other media without making anybody upset about those changes. Nevertheless such « massive » changes may occur. We have seen, here in Paris-France within ten years, the build of hundreds of kilometers of cycling lane and a significative shrink of car space in urban areas. Thermal vehicles will also be progressively banish from the town. This urban transformation makes many motorist angry but the belief in global warming leads to some disrupting decisions. A similar voluntarism could lead to the emergence of specific infrastructures for self driving cars if it appears they could reduce significantly the number of wounded, death, car crash and insurance premiums.
    The second point that let me imagine a change in transportation are « drones ». This is a mature technology for delivering pack from point to point. Its electric, flying and autonomous. Don’t you think that allowing a flow of such objects above car lanes could not be a quite smooth way of developing an alternative transportation/delivery mode that will not make other users getting upset ? Flying cars could progressively merge into this flow…
    I admit all this are guesses and not science but it’s essence of predictions.

    1. Yes, some very good points.

      I had been planning on writing a third post on self driving cars, and how they will change cities. I had planned on using Paris as an example of a city that had been actively changing its traffic patterns recently. And perhaps Barcelona too, though I have not seen that in person, as I have in Paris.

      So, indeed, change is possible. Whether the rest of the world can be as progressive as Paris is another question.

      And yes, we are talking about predictions here. As I think Niels Bohr said, it is very hard to make predictions, especially about the future. I’m sure I will be wrong about many things, and others will be right. But at least I think you and I agree that for self driving cars it is not going to be simple one for one substitution of a self driving car for a human driven car. Instead I think we agree that there will be other changes in the world in order to accommodate the new entities. The discussion here is just about how fast that will happen.

      All best! –Rod Brooks

  14. i can’t be sure whether i am right or not on asking this.
    I saw the your comment on this blog : Merging robotics with human subject is related to some project of DARPA. In that part, I have been curious about how the future BCI for healthy individuals will be. I hope to listen your opinion about the future BCI technology. I made bold to ask. could i get the answer?

    1. I am really out of the loop on this one. I wrote about this in my 2002 book “Flesh and Machines” but I have not followed progress closely since then. Christof Koch is the expert.

  15. I believe self driving cars might be a lot further off than we believe.

    This article points out that the current best-in-class systems depend on a bunch of first world assumptions around roads, rules and weather.

    But I do notice that your predictions are US-biassed, so that is a large constraint on the problem space. However, the things you point out that are known solutions (wipers, tires, wheels, etc) are global in scope. We’ll have to see how these competitive tensions play out.

    1. I admit to being US biased in my predictions as I am mostly debating people in the US (where I live) about the future here. That is not to say that other countries may not do things earlier than the US. Certainly that has been the case with driverless trains (e.g., see the Toulouse subway system; there are no driverless subway systems in the US).

      1. Honolulu may get one in a couple years:

        Vancouver, BC has one that is very close to the US:

        I believe the Miami Metromover is driverless, though it isn’t really a train:

        And of course, various airports in the US have driverless trains bringing people between the terminals.

        (I think Miami and the airport systems go back as far as the driverless systems in other countries, about 30 years, and the general lack of uptake elsewhere in the United States is just due to lack of funding for totally new totally grade-separated transit lines, and not specifically because of slow adoption of the driverless technologies.)

      2. The Honolulu system will be the first significant US commuter system that is driverless. There are about 15 driverless train systems in the US now, but most are in airports, with just about three outside airports. None of them have as much as five miles of track (including the Miami system). Kudos to Vancouver and Canada for their system!

  16. Mr. Brooks, I really enjoyed your essay as I always do. I have two comments/questions which I hope are interesting and it would made my day to read your thoughts on them.

    Consumer (personal) robots. “A robot that can navigate around just about any US home, with its steps, its clutter, its narrow pathways between furniture, etc.” Do you think it’s likely that personal robots will start off being anthropomorphic? (if that was implied). If we assume ‘generation 1’ personal robots will be household only, there’s no need for them to be mobile beyond your house. Hence they might forgo legs/wheels altogether. I’m working on such a prototype. The hardest problem(tech aside) is that first versions (5-10 years) will not be a ‘good’ product. I keep investigating history of PCs during the 70s before VisiCalc, but I can’t understand why somebody would buy those. Was it the cool factor? I have a computer, I’m smarter than you? I bought a computer for my kid, he/she is set for future? Is there a similar motivation to buy a personal robot? Also given the media/entertainment environment in the US where robots will take all our jobs and then kill us, will US be at the forefront? Japanese media is completely different. Anime has been conditioning people since 60s that robots are ‘good’.

    Space. I keep wondering why we assume Mars is the primary target. For space era to kick off, there probably has to be a war. A war in space in this case. US has been an unchallenged hegemony for a while. It controls all seas, there’s no contender at all here. And most likely, there won’t be. The only way contenders can undermine US’s total dominance is to go where there’s nothing yet – space. The best way for a government to protect its space assets/weapons is to have a base on the Moon. Hence I think space era will initially be heavily skewed towards Moon. Also, as you’ve pointed out somewhere in prior essays, the iPhones we’re carrying are products of government inventions put together by entrepreneurs.


    1. I do not think that personal robots will be anthropomorphic at all. The 20M+ personal robots so far delivered by iRobot (of which I was a co-founder) that clean floors are not at all anthropomorphic. I do not expect that any of the elder care robots that will be developed over the next 20 years to physically help people in their homes will themselves have anything like a human form.

      It is still an open question on where the greatest accelerations of efforts in space will be; the Moon or Mars. The US government recently came down on the side of the Moon first, and then Mars. SpaceX talks about Mars, bypassing the Moon.

  17. Rodney,
    A couple questions and a comment:
    How do you, or we, define dexterous hand? It seems we have to agree on some test before we can make a prediction about when they will be generally available.
    It is a bit entertaining that you chose to use the terms “driverless cars” and “self driving cars”, and not “autonomous vehicles”? It took awhile, but we did eventually get past “horseless carriages”. 🙂
    Why do you completely leave out the cost of development. Cost always has a huge impact on the time it takes to develop and produce a new technology.
    No Hyperloop, huh? Bummer.

    1. Oh, Paul, you are being so pedantic. I’ll know a dexterous robot when I shake hands with it!

      As for getting past “horseless carriages”, not necessarily so. I just saw a license plate of a horseless carriage on a moving vehicle in Hawaii over the weekend. Google “hawaii horseless carriage license plate” to see examples.

  18. “Next Big Thing” No Earlier Than 2023? Oh yeah? Then Mentifex here is one of the small furry creatures nibbling away in the footsteps of the dinosaurs. On New Year’s Eve I took my Acer netbook with me to a Starbucks coffee shop to code some AI. For eight hours I sat there coding AI while people milled around. The next day on 2018-01-01 I changed the title of to “AI Mind Maintainer” and the title of to “AI Mind Maintainers” in the plural. On 2018-01-02 I went to a Seattle coffee shop to tell my buddy that I was going to concentrate on my AI project and not initiate conversations with coed beauties, but I offered to buy coffee for the U.S. Marines lieutenant standing behind me, and she and I spent the next eight hours together making a podcast that included my AI project. “There will be many claims on this title earlier than 2023, but none of them will pan out.” The race is on, Dr. Brooks, the glove is thrown, the die is cast, and I will see you at the Singularity. -Arthur

  19. Dear Rodney
    Great essay. I agree with most of it. However, I wonder about the relevance of the listed technologies in the future. There could be much larger shifts in our society that make the predictions meaningless. AI, big data and our connected world have such a strong social and political impact (already today!) that we might not care so much about future cars, robots and space travel. Our society might develop into directions that will change our views and shared values more than we can imagine. It is right that cars, planes and mobile communication technologies have not changed dramatically within the last 30 years, as Donald stated above. However, our behaviors and attitudes have changed. Watch our children, how they communicate and meet with their friends. See how the smart phone determines the daily pace of family life (even at the restaurant table). Our daily work methods have changed – not many office workers have used a PC in 1988. See how big data and IOT are leading to novel methods (e.g. blockchain technologies) that will have an effect in our daily life. On a large scale, see how the transparency of the internet and “real news” can affect totalitarian systems (arab spring revolution starting in 2010 in Tunesia) but also how “fake news” can harm democracies. Thus, I wonder, whether we will care about cars, robots and spaceships in 2050. Or to formulate it a bit more drastically: will our societies, our democracies, our families be strong enough to survive this storm of changes?

    1. I agree that there could be lots of things that overwhelm these technologies. But this is my best shot at what we can see now. I doubt that any of the world shattering events will speed up any of these innovations. Let’s hope, and work towards, there are no world shattering events.

    2. We are a lot closer to conditions that might cause a global famine than most people suspect, or would even believe in most cases. For example, about forty percent of the world’s surface has turned to desert in the last forty years, and there are now quite credible estimates that topsoil erosion will make farming in Great Britain impossible in about sixty years. And then there is the rapid loss of biodiversity in our agricultural crops that leaves increasing fractions of our food supply vulnerable to just a few pests. Even now the world supply of Cavendish bananas is being wiped out by a fungus that has so far proved unstoppable. We can manage without that variety of bananas, but what happens if we lose millions of acres of wheat or rice in just a few years? We are also losing a great many of our wild pollinators, upon which we are utterly dependent. No pollination, no crop.

      This is just “boring old low-tech farming”, but if we get it wrong then we will finally have an indisputable use case for driverless cars. Problem is, they will also be passengerless.

      In short, the high-tech world is completely predicated upon a natural world that may not exist in fifty to a hundred years.

  20. I am a layman – 70 and recently retired. I do have a brother and a son who are both alumni of the august university that you belong to – the closest I am to illustrious thinkers!!
    I truly appreciated the thinking and the logic behind your reasoning and conclusions.
    I am amazed that such concepts could be written about in such lucid language and people like me can appreciate and comment on them!!
    I make bold to suggest that I do feel that there is an element of scepticism – or pessimism – in the article about future developments. – this is based on what I have seen happen in the last 60 years or so; that is ever since the ‘man to moon’ program was launched
    Anyways many thanks

  21. A minor nitpick:

    “containing capsules that are accelerated by external air pressure”

    The propulsion in current Hypeloop prototypes is a form of maglev. In that respect, the system is not similar to the pneumatic ones attempted a century ago. Elon Musk’s original white paper on Hyperloop proposed using a linear electric motor for propulsion and a turbine on the capsule to compress what little air remains in the near-vacuum tube to form an air cushion under the capsule for levitation. I believe that was deemed unworkable and the prototype designers went for maglev. That’s (just) one of the reasons why Musk’s cost estimates in that first white paper were wildly optimistic: he was assuming a dumb tube with not much infrastructure outside of the linear motor segments, but the current prototypes have a maglev track that needs to be aligned extremely carefully for high speeds, canting etc.

    1. I have not followed the details as well as you have. But clearly such a major shift in strategy indicates that the technology is not yet developed, so I feel confident in my NIML prediction.

      1. Agreed. To my mind, the major problem with Hyperloop is capacity. I can imagine a prototype can be made operational for rudimentary passenger travel trials, even in the relatively near future. But Hyperloop was billed as a high-speed rail killer, with Musk’s disparaging comments about the Calfornia HSR project. To reach HSR-like capacities (say 1000-passenger trains every 20 minutes in one direction), you’d need pods of 20 or 30 passengers traveling with headways of less than a minute. The various white papers on Hyperloop claim that this can be done at speeds approaching the speed of sound. They also understate the current, already operational capacity of HSR lines in e.g. Japan and France in order to make Hyperloop look competitive.

        I simply don’t buy the claim about the headway. A pod traveling at the speed of sound in a tube, followed by another one 30 seconds behind, and another one behind that? What could possibly go wrong? The original white paper basically waved away the safety concerns, stating that the pods will simply naturally come to a graceful stop should the vacuum fail, and in case of pod failure, an automatic brake would be triggered on the whole line. To this day, I’ve seen no real explanation of how people would be evacuated from the tube. And if you require some kind of a safe headway, one that would allow for braking from the speed of sound without killing the passengers, the capacity claims go out the window.

      2. Such a major shift in strategy sounds like a completely different technology, IMO. One that is piggybacking on the “Hyperloop” name merely as a way of gaining publicity. What a farce.

  22. My hat in the ring of the next big AI thing is: agorics for guiding software modification/experimentation.

    Agorics is a thing from the late 80s where programs bid for computational resources using a currency.

    This can drive software evolution.

    My own take on in goes towards the Koch end of things. I want humans to be the source of currency based on what we think is good. It hope that we can think of the market as an extension of our dopamine/neutrotransmitter system. And the computers as a proper extension of our brains.

    My current progress is just a demo. A VM that can run arbitrary code, but uses a market and feedback to determine which of two programs gets to output to the text stream. Details here

    The CPU/memory auctioning is there as well, so it is moderately complete from an agoric point of view. Very deficient in many other respects, documentation etc.

    No academic papers as yet (some blog posts around though), as I am not in academia, so don’t feel the need. Time-lines, who can say. It is going to take us a long time to figure out how to work with this kind of system. I think I’m going to work on the security/computer architecture/tooling aspects of this for quite a while (FPGAs have always interested me). So it might take a while for it to light the AI world on fire.

    It is actually mildly nostalgic to see that Mentifex is still around.

  23. “A robot that can carry out the last 10 yards of delivery, getting from a vehicle into a house and putting the package inside the front door.”

    Ahti Heinla, one of the founding engineers of Skype is working on it for 4+y. Starship is doing last mile delivery as public pilot, constantly improved for a couple of years. That is based on experience from NASA challenge, robot kuukulgur. They have driven 100+k km and are delivering packages and pizzas as commercial pilot. They run at least in Tallinn, London, Washington DC and other cities.
    Task is changed from fully autonomous to economically viable last mile delivery, mostly autonomous using very cheap sensors and cheap processors. They use operator remote control in difficult situation like start a road crossing, riding an elevator or not mapped / changed area. I remember that they were proud of being able to find precise location with very cheap, not depth cameras.
    Found a 1y old video:

    1. Yes, I’ve been watching that effort for a couple of years now. For the really elderly, or otherwise, who can not carry their packages up whatever steps there are to the front door, this solution still leaves a problem. But for lots and lots of people in lots and lots of geographies this will be very useful.

  24. Well done, Rodney. I agree with your timing on AI and robot stuff. However, I think your sequencing is more important than the dates, as the sequence of inventions — what comes first, which enables the second, how the third impacts the first — is more informative than the dates. The fault of almost all science fiction is that the cool technology in their worlds don’t have precursors. There is suddenly hyperdrive, time warp, or flying cars without the precursor technology from 10 years earlier, or 30 years earlier, when it half-worked. And these precursor species form the ecosystem for the new novel species. The old rarely leaves. In San Francisco today there are Teslas plus thousands of foot-powered bicycles. The subtext to your moderation of expectations is that there is a lot of precursor technology required to make the charismatic technology come true.

    1. Thanks Kevin, I really appreciate your remarks. And yes, you are quite right to point out the sequencing; that was definitely part of my analysis that led to the dates that I gave. I trust that all is going well for you!

  25. why are flying cars such a hot topic? what’s the benefit? i rather see the automated condensed traffic much more convenient, economic, etc. including an automated “pick me up in half an hour” functionality rather than “this is MY flying car, and I personally can (ram it into my boss’s office window) enjoy the (complexity of 3D navigation with all its controls) view”.

    1. I do not think flying cars should be such a hot topic. BUT, a Chinese auto maker recently bought a Massachusetts based flying car company, there are at least two Silicon Valley based flying car companies, and a year or so ago Uber was claiming that they would have autonomous flying Ubers by about 2020 or 2021. So I put flying cars in my predictions as I believe that they are still total overreach.

  26. Nice list! I find the timelines a little dispiriting but also realistic. A sign of good compromise between optimism and practicality I guess.

    You’ve encouraged me to update a prediction I made some time ago about when new construction American homes would start offering “No kitchen” or “Minimal kitchen” options. My theory was that the automation of food supply chains, restaurant kitchens, and self-driving delivery would make food delivery cheaper than buying groceries and cooking for yourself. And eventually this logic would change consumer preferences in home construction. I’ve seen progress towards this scenario but as you keep saying, the progress has come more slowly than I initially expected.

    The automation of supply chains continues apace (Amazon drives this, especially with their purchase of Whole Foods) and the progress towards self-driving and “curb-to-doormat” delivery is addressed in your post above. Also, apps like Uber Eats are changing the restaurant industry to drive the creation of “delivery only” restaurants in industrial parts of town where the rent is lower. What I’ve been surprised by though is how the few fully automated restaurants out there like the ones powered by Momentum Machines haven’t expanded faster. The cashiers are getting replaced faster than the kitchen staff, with deployment of tablet-based ordering kiosks. I guess there are more kinks to work out than apparent on the surface.

    The full scenario (from November 2015) is linked in my signature here.

  27. I really cannot see any reason for us to send humans into space. With robotics developing at such a fast pace (my observation) they will be able to do all the mining and transporting and whatever, in a very short time. It seems to me that sending “a man to Mars” is nothing but antiquated ego. We should turn around and look at our planet and do whatever it takes to keep it from making us extinct.

    1. Some people argue that having humans as a two planet species is a safety strategy in case we destroy ourselves on Earth.

      1. “Some people argue that having humans as a two planet species is a safety strategy in case we destroy ourselves on Earth.”

        Life on Mars is not hospitable to life, at least in the form we take now. How would humans survive there, when here on earth even with all the resources we have, seems unsustainable? Why Mars? Some people claim many things, others look for earth-like planets which may have similar variables as Earth which is equally unfeasible due to long distances..

        ” We should turn around and look at our planet and do whatever it takes to keep it from making us extinct.”
        I agree with you Michael here but AI is a way more cool and interesting subject while thinking about and taking care of Earth is for wimps.

  28. I think a common mistake made by past futurists is that a lot of technologists simply fail to consider the *economics* of their ideas. That is to say, there’s lots of cool possible technologies, but what is their actual *economic* value?

    Predictions of flying cars and hyper-loop are good examples of the above – cool ideas with little to no economic justification. What is the actual *economic* value of whizzing people around at ultra-high speeds? Not clear at all. These technologies , whilst ‘cool’, would be complex, expensive and environmentally destructive. There’s a strong analogy here to supersonic flight. Yes, Concorde was cool, but ultimately had no economic justification.

    The next ‘big thing’ for AI? I think you’re entirely right that it’ll be a ‘blast from the past’ – something people have already been working on for a long time, but the potential of which hasn’t yet quite been recognized.

    My pick? Fuzzy Temporal Logic for NLP ( Natural Language Processing). Click here:

    I’m picking a return to a logic-based (symbolic) approach, but this time combining several different strands of non-classical logic (modal, fuzzy, temporal).

    Watch closely for increasing chatter about ‘Reflective Symbolic Models’ and ‘Many-Valued Logic For NLP’

  29. Great essay and BRAVO for putting dates on things!

    Easy clarification question: do you mean by “electric cars” pure battery electric (like Bolt) or do you include plug-in hybrids (like Volt)? I know there is mass confusion about this, as the press and the public often do not see the blurred lines between “electrified” and “electric.” (Thus in the Volvo pledge to be 100% “electrified” by 20XX the definition of the term is so broad that all of Lexus’s European models ALREADY meet that criterion, since Volvo was counting mild and 48V hybrids).

      1. I agree re gasoline, I was only seeking clarity on the electric term. Just as when I say “gasoline” I also mean “diesel,” which is still in wide use in Europe. The distinction matters since one can see forecasters such as yourself, UBS, IEA, etc. all coming up with forecasts, but one is counting BEV (as you are), one is counting “plug-ins” (BEV+PHEV) and one is counting both those and strong hybrids (HEVs). Gets confusing.

  30. I think the package delivery use case of driverless cars is one factor that will prove very different from existing car use. I am sure it has been said before – but one of the unintended consequences of autonomous vehicles will be more traffic congestion, because people will be ordering cables, dinners, underwear and having each item delivered by an autonomous vehicle. We really need hyperloops for package delivery – not cars.

    1. Yes, our streets are already getting crowded by delivery vehicles. In my neighborhood in Cambridgeport, in Cambridge, MA, most streets are essentially one lane one way, and now even on Sundays they are often blocked by delivery vehicles.

  31. Once the dust settles with the AI hype I think we will find that there is quite a lot of low hanging fruit to pick with basic AI systems working in combo with humans. Picking actual fruit would be a perfect example most likely not practical by robots before deep learning and now easily within reach with what we have already.

    1. I am happy for you that you have found a secret path to AGI. There are thousands of others out there with their own secret paths to AGI. We’ll see who, if any, wins.

  32. Rodney, been watching your robotics (and AI) work for a while.
    As someone who has been involved in industrial AI applications, it’s refreshing to come across a realist.

    DARPA feels that the next evolution in AI is the ability to automatically generate abstractions from a set of instances followed by multi-initiative problem solving (automatic application of multiple techniques to deal with a problem). IYHO, what do you think ?

    1. There are so many ideas out there, and that is good. Most of them will lead nowhere. But sometimes, e.g., with Deep Learning, a technique blossoms and is incredibly powerful, so we need thousands of people trying ideas to find out what ones might work. Experience tells me that evaluation before hand (even for thirty years as in the case of convolutional neural networks) gives no real prediction of success. So, no one knows what will be successful right now, and at some point it will become apparent that something is successful. But AGI will require thousands of such successful techniques, so it is a long way off.

  33. This is a great post and I love that you have been so precise about your predictions. Is there a way to translate these into numbers? I.e. would you say that the date that you assign is something like the date at which you have 50% credence that the event will happen by that date? Or some other number? It would be interesting to turn some of these into issues on a forecasting platform that I’m involved in, and see to what level others quantitatively agree or disagree.

    1. I tried to give “exact” dates by most of them being “No earlier than “. So it is not saying that things will occur in the named year, just that I am 100% sure that they will not happen before that. That way it will be relatively easy to score me as to whether I was right or wrong.

  34. Really enjoyed your article. You make good points about how sociological and political factors will slow down some of the progress touted by researchers and engineers.

    While they look only at the technological aspects (and are still quite optimistic about those, for many, very human reasons), those who are outside these “bubbles” see things differently and have real-world, valid concerns about the impact of these changes.

    i think most of us want to see these technological innovations, but change happens at its own pace.

  35. When you made these dated predictions, did you take into account China’s nationalistic fulled desire to be the sole AI superpower? Since they are punching out hundreds of AI Researchers at breakneck speed a year, how do these affect your timelines? How will China’s AI push affect AGI timelines?

    1. Yes. And two points. 1) “The Mythical Man Month” 2) As we have seen many times before, it takes a formalization of what the problem is (e.g., the Leonard/Durrant-Whyte formalization of SLAM) along with thousands of eager researchers to make a difference. In my predictions I did not say who would come up with the critical ideas, only when they would get adopted. I feel very comfortable on all my predictions, and only worry that I used too many BY’s rather than NETL’s.

      1. P.S I love the post.
        Consider this.
        Worldwide AI Arms Race = More Researchers Worldwide.
        More AI Researchers mean a reduction of Timelines, would it not? There would be more potential of the “Eureka” Moment from one of these researchers. I feel like timelines would be widely out of flux all the time.

  36. Also when I went to dig into “Leonard/Durrant-Whyte formalization of SLAM” I found nothing. Care to explain this?

      1. I’m with Jeremy, when the U.S Government funded research, breakthroughs followed. (GPS and the internet are two that come to mind) Once governments throw money at technology, innovation seems to follow. Is research linear or exponential? And who’s to say once China bites down on the “AGI Problem” These timelines go down the drain? 1) “The Mythical Man Month” only seems to address an ‘ongoing’ project adding resources to it. This does not address the small “new” projects (Baidu, Tenecent, ect) that are starting up. “2) As we have seen many times before, it takes a formalization of what the problem is”. As more small projects start up (with AGI being the goal) all it takes is one meritocratic researcher to “formalize the problem” and its solution. Unless more money + researchers actually diminish innovation which seems unlikely.

      2. We’ll see what happens! I have dated my predictions with BY, NET, and NIML. Every year I will post an update on which ones I got right and wrong as time advances. Tune in on January 1st, 2019 for the first update.

  37. Great points Rodney. I have long thought about similar things, i.e., how long technology actually takes. Another way to look at it is: How long does old technology last? Edison’s light bulb is a good example. It’s scale of economy, (and the intrinsic laziness of technology) made for a 120+ year dominance before LED technology really took it out permanently, and not without legal help by the way! This blog really got me thinking about my own inventions in modular arithmetic, and yes, even if you’ve got the best thing since sliced bread, it doesn’t mean anybody is hungry!

  38. My growing suspicion is that we won’t really see “flying cars” as a major social phenomenon in the short term. Rather, we’ll see more and more granular air transportation services. Those aircraft will not be cars, they’ll be aircraft, but they will be smaller, autonomous, and much more plentiful than current air transport services.

  39. Very interesting insights, thank you very much. The only one that looks completely unrealistic to me is the first manned mission to Mars by 2032. I’d put it NET 2050, most probably by 2069±1y (± because I haven’t check the most favorable launch windows in the case they want to use Hohmann transfer orbits, it may be launched 2068 to land 2069, or launched 2069 to land 2070). Obviously, to celebrate the flight of Appolo 11.

    See, for example:

Comment on this

Your email address will not be published. Required fields are marked *