Arnold Kling  

The Technofuture

Me in The Economist... Pre-Gloat...

Christian H. Nesheim asks folks which technology will have a big impact over the next twenty years. A sample response:

But from an ethical perspective, I think the single most important technological development will be the mass-production and consumption of in vitro meat. ..The global adoption of a cruelty-free diet will mark a major evolutionary transition in the development of civilisation.

But will the EU allow something that is not "organic"* to drive its farmers out of business?

Anyway, a lot of the respondents seem to think that artificial intelligence will be here real soon, now. I remain a near-term bear on that front. I think synthetic biology is the technology most likely to break through over the next twenty years.

I look at the problem this way. To get something useful in the nanotech world, you need to be able to accomplish both self-replication and being able to control what the stuff does. With "dry" nanotech, we might be able to control what the stuff does, but self-replication is hard. With "wet" nanotech, self-replication is easier, but controlling what the stuff does is hard. Not really knowing anything, my instinct is that it will be easier to figure out how to get living things to do stuff that we want than it will be to get things that do what we want to replicate and construct themselves. So, over a twenty year horizon, my "money" is on the wet nanotech guys (actually, I don't have any investments in the field.)

Comments and Sharing

COMMENTS (9 to date)
Michael Keenan writes:

My understanding is that there are plenty of "dry" nanotech applications that don't require self-replication, and in fact the ability to self-replicate is probably undesirable unless required. Nanobots could be mass-produced in factories. The mass-produced nanobots could presumably do whatever it is we think nanobots will do: measure or remove pollutants, kill germs and cancer cells, carry oxygen through our blood, espionage, clean up broken nanobots, etc. We'd just need to regularly send freshly-produced nanobots wherever they need to be, to replace losses.

SB7 writes:

@Keenan, a lot of what is called nanotech in the "dry" realm these days is really materials science and engineering, rather than individually functional bot-like mechanisms. I'm not optimistic about those bot-like applications making a big splash in the next couple of decades.

~ ~ ~

"Anyway, a lot of the respondents seem to think that artificial intelligence will be here real soon, now. I remain a near-term bear on that front."

Agreed. I'm in AI research, and I think people massively overestimate how much impact AI will have in the short term. (Though they also underestimate it long-term, like most technology.) Literally every time in the past two years I've told people I work in AI they make a joke about robot uprisings enslaving humanity. If they only knew that we can't even get machines to load a consumer dishwasher they would be less concerned about them conquering the world.

If there's one place AI could make a big difference in the next 20 years, it's automated driving, but the adoption of that tech is going to be more limited by safety regulations than it does on the technology itself.

Felix writes:

Free range computers.

Auto-driving is one example. Auto-dishwasher loader would be another. Auto lettuce-picker, another. Auto-machine fixer, another.

The 2020's seem like a good time for free range computers to go up the S curve.

CPU cycles will be 1/100th to 1/10000th the current price during that time period. Communications will make cheap, remote CPU cycles available for some serious number crunching in that dishwasher helper.

Required: ways for non-AI gurus to put together such beasts. Computer "brains" don't seem to be the problem. Rather, the problem seems to be sensor processing - figuring out what's important of what the machine's "eyes" see. Solve that problem and let Moore's law and world-wide shipping take care of the rest.

But 2020's? If we're on track for $1000, full DNA sequencing by 2015, then 10 more years seems like enough time to find ways to exploit that in interesting ways.

Government? Governments are structured for a pre-Internet world. And peak-nation-state happened in the 60's. The odds are good that superior forms will evolve by 2030. Won't have much day-to-day effect, though. Sigh. :)

MernaMoose writes:

In vitro meat will change the world? Um, okay. Whatever floats your boat. I agree this is in the right vein, but test tube groceries are going to be the biggest news over the next two decades?

So what is going to be the big news over the next two decades? Let's see.....

AI? I have to agree with SB7 and Arnold, nothing there in the near term. We will have AI when we figure out how to replicate a bug's brain with a computer. The brain of a very simple, very stupid bug would do nicely for turning the world upside down.

Energy? It's really not looking good over the next two decades.

Nano? Lots of smoke but not so much fire. I can remember not long ago when people were claiming MEMs (micro-electromechanical devices) and nano materials would revolutionize everything. An over sell.

Nonetheless, I believe advances in materials will be significant over the next two decades, even if it's not the top of list.

Robotics? Still waiting for Energy and AI to make some major advances. At least, if you're talking about robotics in the sci-fi novel sense.

Space? Also on hold, largely for the sake of Energy and its near cousin, Propulsion. Not to mention some semi-major materials problems.

My money goes down on two items being tied for first place:

1) Big bio-med break throughs seem quite probable, even in 10 years.

2) All the things computers are going to change that we haven't even dreamed of yet. I suppose you could call this the Pre-Robotics stage.

Felix writes:

MernaMoose: Materials

Agree. I subscribed to Science News for some 30 years. Month after month, year after year there were articles about some breakthrough in the general area of "materials". Always slightly below the radar, but a lot of 'em appearing in a product near you today.

The materials guys get no respect, no notice. But, pound for pound, in daily living, they keep changing the world as much as anyone.

Remember when straws would collapse if you sucked anywhere near hard? Now think of 100 things similar. No problem, except these changes are so overlookable.

Jeremy, Alabama writes:

There will not be an AI "breakthrough" that is recognizable as such - no sudden synthetic consciousness.

But "intelligence" is more organizational than inspirational. The internet, GPS, satellite imagery, webcams, blogging and tweeting, google, facebook, babelfish, XML. These technologies are organizing data on a planetary scale. AI apps that affect people's lives tend to exploit these, rather than some new neural net algorithm from basic research.

AI's next wave of applications will need to penetrate into humanity's error-filled and heavily-litigated actual reality. Driving cars, for instance. It is this universe of lawyers, rather than software nerds, that will slow down Kurzweil's Singularity.

Peter Finch writes:

I'd vote "space".

Between XCOR, Armadillo, Masten, SpaceX, and others, a real renaissance is underway. It seems that serious space access technology can now be developed by a small team for a few million dollars. I predict by 2030 that more mass will go to space on vehicles built by new firms than by Lockmart or Boeing, and that prices will be falling rapidly. Large numbers of people will question why we have a NASA.

I agree with the posters above regarding AI and robots. We're nowhere close. That's the technology of 2130.

Matt writes:

Self-driving automobiles seem like a good candidate for something that is both likely to happen and likely to be fairly transformative.

The potential benefits are pretty huge. Self driving vehicles, that can communicate directly with each other, can potentially make better use of the road, resulting in fewer accidents and fewer traffic jams. Additionally, commute times are no longer wasted time. This would likely lead to more spread out cities, and lower property values near city centers. Lastly, self-driving automobiles can still be in use even when the driver is otherwise occupied. This could be a huge boost in utility, and also lower demand for support infrastructure, such as parking spaces in high property value areas.

The technology is already at a fairly advanced stage. So I think that if self-driving automobiles don't become a reality, it's likely to be due to factors other than technical limitations, such as peoples' overconfidence in their own driving abilities, companies' aversions to lawsuits for accidents, and interest groups, such as truckers or taxi drivers, pushing for legislation to limit applications.

Prakash writes:

In the light of Romer's "rules are technology", I would like to propose - better utilisation of the crowdsourcing phenomena.

Groupon, The Point, and similar such contingent contract based sites will allow people to come together at a much lower cost than before. This will enable many marginal phenomena which could not have existed before - eg. carsharing, printer sharing or projector sharing.

i know it's weird to think of such capital conserving things in a capital surplus time like this one. But this capital surplus is not a reality for the majority of human beings.

In short Bottom-of-pyramid sharing phenomena.

Comments for this entry have been closed
Return to top