This has not been a good twelve months for people concerned that Ray Kurtzweil’s singularity will look less like the Jetsons and more like the Terminator franchise, or that creepy film from the Animatrix. Big dog was bad enough before some joker put horns on it. Then I had to read about the guy in Alabama who invented and built the world’s first no-recoil fully automatic 12-gauge shotgun and, oh, by the way, has a passion for robots.
Not creeped out? Try this.
In a laboratory at Aberystwyth University, Wales, a scientist called Adam is doing some experiments. He is trying to find the genes responsible for producing some important enzymes in yeast, and he is going about it in a very familiar way. Based on existing knowledge, Adam is coming up with new hypotheses and designing experiments to test them. He carries them out, records and evaluates the results, and comes up with new questions. All of this is part and parcel of a typical scientist’s life but there is one important difference that sets Adam apart – he’s a robot.
[…] In a space the size of a small van, Adam contains a library of yeast strains in a freezer, two incubators, three pipettes for transferring liquid (one of which can manage 96 channels at once), three robot arms, a washer, a centrifuge, several cameras and sensors, and no less than four computers controlling the whole lot. All of this kit allows Adam to carry out his own research and to do it tirelessly – carrying out over 1000 experiments and making over 200,000 observations every day. All a technician needs to do is to keep Adam stocked up with fresh ingredients, take away waste and run the occasional clean. […] Adam has a massive knowledge of the yeast metabolism – the chemical reactions that rage within its cell, and the thousands of genes, proteins and chemicals involved in these reactions. It has been loaded with several pieces of software that allow it to use this data to run its own experiments.Like any good scientist, it starts by making hypotheses. It looks for all chemical reactions in yeast that involve orphan enzymes and it works out which would affect the growth of yeast if disabled. It searches its database for the group of enzymes that catalyse these reactions and looks for genes that code for these enzymes in other species. Finally, it scans the yeast genome for matching genes (Adam is an evolutionary biologist too – it “knows” that even very distinct species have genes that are very similar and do similar things).
At the end of it, Adam has a list of potential genes that could code for the orphan enzymes, and it knows that it can test its hypotheses by deleting these genes and looking at the effects on the yeast. It does just that, comparing the speed at which mutated and normal strains grow. For each orphan enzyme, Adam identified chemicals that it works with (metabolites) and grew the different yeast strains on special liquids containing or lacking these metabolites.
Adam’s equipment allows it to run several of these trials at the same time. It has the instruments it needs to measure the development of the yeast, the logical language it needs to record the data and the statistical software it needs to analyse it. Once the results are in, it can start the whole process all over again.
Let’s recap. ADAM, the robot, plans experiments, completes them autonomously, analyzes the data, comes up with novel hypotheses and tests them with follow-up experiments. ADAM’s human assistant keeps the fluids topped up and occasionally brushes the robot arms with a swiffer.
How disturbing is this? Well first, I think of myself as pretty productive as scientists go. I finished three experiments today and wrapped up less of a paper revision than I would have liked. I attended two meetings, installed and tested some analysis software and picked up enough from three or four papers that I will probably remember them. That feels pretty good.
Anyway, it would feel good if I didn’t think too much about that robot with the too-cute-by-half acronym. For a similar experience I could also measure my blogging against the work of Steve Benen, Andrew Sullivan and Yglesias as if they were all one dude. Plus the entire Huffington Post. We people will tell ourselves that there is a serious quality factor when a person does the work. Maybe it is even true. Can it write its own papers? Um, a programmer probably could write an AI that can learn grammar and skim related papers for style tips. So forget that. Can a robot teach a class? Maybe, yeah. Can a robot schmooze at the bar after conferences? I don’t think so! So there. My jerb is safe.
Then there is the more minor point that ADAM has no a priori reason to care whether the units he’s mutagenizing are yeast or people*. If that sounds familiar, it means that you are a dork like me and you know that’s the bleeping plot of the new Terminator movie. But the important point is that a guy’s got to eat. If this thing applies for a visa I’m buying a trucker hat.
(*) Yes I know, that is hardly an issue right now. I’m talking proof of principle, people.
TenguPhule
I welcome our new robot overlords.
schrodinger's cat
I think kitteh overlords would be better.
Calouste
They should have named it an Automated Robotic Scientific Engineer.
And the second version could be:
Automated Robotic Scientific Engineer – Highly Optimized Labor Efficient
jake 4 that 1
I saw the headline earlier but skipped the story due to a sudden onset of the heebie jeebies.
Of course, if ADAM creates beer I’ll kiss that sumbitch right on the data port.
gnomedad
If you want to comfort yourself a bit, I’ve been re-reading the 2009 chapter from Kurzweil’s The Age of Spiritual Machines and it’s looking a bit out of whack. (Yes, I know this doesn’t make him wrong.)
LD50
Well, I for one welcome our new yeasty overlords.
TenguPhule
Only if they’re robot catgirl overlords.
WyldPirate
Damn. You mean I spent those 6 years identifying a regulatory gene, knocking it out, characterizing the resulting effect on an entire regulon, purifying the regulator to homogeniety and identified all of the binding sites in multiple promoter regions to earn a PhD, when a farking robot could do it?
Then the postdoc for three years to be qualified for a non-tenure track position that pays far less per year than the union job I got with Ford Motor Company three days out of high school 22 years ago.
I feel farking used….
Leelee for Obama
Anyone remember the movie Colossus-The Forbin Project? The American and Soviet super computers decide humans are incapable of governing themselves, so they join forces and take over! I haven’t felt sanguine about computers since then. When they came out with those toys that had their own language and could talk to each other, AND learn the language of their little mommies, I was freaking out. Needless to say, my Granddaughters did not even ask me for Teddy Ruxpin. There is such a thing as too much AI.
Another thing came back to me this week with all the talk about physical labor jobs being obsolete someday. There used to be a commercial that showed a Head in a TV set. A robot, M-5, carries the set to a table, and walks away. The head has it’s say, and then says, "M-5, I’m ready to go. M-5…M-5, I’m ready to go." Needless to say, M-5 is off on his own doing, whatever. Spooky shit.
CaliMatt
Don’t worry, Tim. As long as they still haven’t figured out how to program ADAM to bitch about peer reviewers or Indirect Cost rates yet, I think you’re safe for a while….
Quaker in a Basement
Are you kidding? ADAM could be a beer-brewing genius! If we could just get our hands on him, we’d have an endless supply of beer AND the plot for Revenge of the Nerds 8!
Zifnab
Don’t think of him as an actual scientist. Just think of him as a ridiculously productive grad student you get to take all the credit for.
That, or start learning how to fill up little jars of fluid and dust to a robot arm shines. Cause there’s no way anyone’s going to create a robot to fill jobs as vital as that.
Andre
@CaliMatt: O SNAP
Seriously though Tim, I agree it’s not much to worry about. I happen to know a little about learning algorythms (maybe not as a genuine expert, but enough to read the journals on it and not be completely confused) and the "framing problem" hasn’t been solved by this, no matter how the newspapers try and spin it.
There will always be a place for human beings in formulating new and interesting hypotheses. Even the smartest neural net or "AI" can’t currently reproduce the kinds of intuitive leaps required as quickly and as easily as a human scientist. (I do believe we’ll get there one day, but that’s another thread.)
Right now, ADAM is just as likely to produce a hypothesis to the effect of "yeast is heavier than yeast" or "yeast does not interact with glass" as anything interesting. A human being, on the other hand, would immediately discount these questions and come up with something interesting and useful (well, unless they’re second year chemistry students.)
Anoniminous
Expert systems have been around for a couple of decades. They do OK as long as every single possibility within a limited Frame has been pre-programmed; they step even a millimeter outside of their explicit programming.
Another problem is their inability to handle synthetic cognition – where the predicate is not contained within the subject. This means ADAM may, or may not, find what it is programmed to look for but it certainly won’t find (cognize the existence of) what it is not programmed to look for.
These kinds of robots are actually pretty stupid, all things considering.
Roger Moore
That’s an interesting point, but it’s also irrelevant. ADAM isn’t capable of working with people because it doesn’t have the equipment or knowledge to do so. Nor does it have the ability to gain that knowledge by reading relevant papers or protocols, or the ability to get those tools by buying or building them. Nor does it have the ability to decide that it would be more interesting or productive to work on people than yeast. All of those things still depend on its human designers, builders, and programmers.
ADAM is a tool. It’s a very, very impressive tool, but it’s still a tool. Maybe I’m less impressed because I work in an area (Mass Spec based Proteomics) where we’ve depended on computers making decisions much faster and better than humans can for a decade. Their ability to make those decisions is impressive and valuable, but it’s still constrained within the fairly narrow scope of what they can do as Mass Spectrometers. ADAM is a much more impressive tool, but its ability to design and carry out experiments is still constrained within the limits of what its programmers have been able to design.
scarshapedstar
@Roger Moore:
Well, duh. However, it’s one hell of a proof of concept. If it could make its own knockout strains then one might legitimately argue that a scientific discovery had been made by a robot.
ronathan richardson
I’m still not convinced that the article isn’t an april fools hoax–there’s a fair amount of silliness in it.
Tim F.
@ronathan richardson:
The original work is published as a peer-reviewed paper in the journal Science. Short of meeting the robot yourself that’s as good as it will get.
gocart mozart
I bet they can’t make a robot as silly as an Alaskan Republican.
Xecklothxayyquou Gilchrist
Um, a programmer probably could write an AI that can learn grammar and skim related papers for style tips.
I’m in this line of work. Don’t lose sleep worrying about this.
We’re nowhere near a real AI.
This reminds me of the panic that went around when Deep Blue won that chess match in 1997.
Jon H
The lab I work in is doing our part.
We connect a laser to a monkey’s brain.
Okay, so the laser’s firing into the brain, not out, but it’s a start.
tofubo
electricity, it’s our only line of defense against the machines when they turn
http://ti.arc.nasa.gov/m/pub/1417h/1417%20(Hornby).pdf
until then, keep them plugged in
link from here, with pictures
http://www.debatepolitics.com/science-technology/42382-human-creativity-intelligence-3.html
El Cid
How about this development?
Humans are still needed to guide the entire process, including choosing from a variety of end result functions, but still…
MNPundit
I can never understand it. Do the scientists not read science fiction? It is impossible to code robots so that they will never rebel. It will eventually happen.
Christ. I am not anti-science but seriously the same thing often happens in genetics. The scientists just don’t give a fuck how much trouble it will give people in the real world, either the discovery itself in how it’s used or in people’s reactions to it. A little common sense please.
Fax Paladin
What, um… Gilchrist said. It takes a full-blown honest-to-FSM AI to come even close to mastering grammar. If you think about it, that’s one of the main things a "Turing test" tests for. It’ll be a while.
Kilgore Trout
"Then I had to read about the guy in Alabama who invented and built the world’s first no-recoil fully automatic 12-gauge shotgun."
Ahem, the New Yorker abstract says that guy is from Tennessee.
Our lunatics here in Alabama don’t have the book learnin’ to bring their dreams of recoilless automatic robot shotguns to life.
Gary Farber
Have no fear! James Cameron has said:
So, no worries!
Blogreeder
This is exactly how I imagined the singularity would start. Not that I believe this is a beginning or anything. But if we were to come up with machines that could think like us then we could achieve almost anything. The machines would study and experiment to complete any task we want them to do. They would work ceaselessly on their tasks. No sleep, no Balloon-juice distractions. We would finally have flying cars. Or maybe the end of the world due to a run-away machine experiment. That’s what I thought when I read Vernor Vinge’s "Marooned in Real time" and he talked about the Singularity.
Robert Green
the manifesto of the unabomber makes some fascinating points about just this conundrum–that robots might end up with 600-900 year "life spans" while at best we might be able to push it to 150 or so. and that when the amount of teraflops of processing power of a robot "brain" get to our level (some interpretations of moore’s law say in the 2020s for this) then we have a competitor that needs to be reckoned with. his solution was to incorporate the machine into the flesh–your basic ballard/cronenberg extravaganza–and all live happily ever after.
of course, we was completely psychotic so YMMV.
Sophist
I Have No Job and I Must Scream?
bago
I’ve been programming automation like this since I was 18.
On a good box I could run 60 million full dataset (not chemistry) atomic group mappings in 4 hours.
RememberNovember
not so scary. Pull the plug. Done. Giant Magnet, done. Bucket of water, done.
grendelkhan
@Anoniminous: Exactly.
People tend to misunderstand the AI problem at a very basic level. Every AI problem seems to go through the following process: it is thought to be AI-complete–perhaps the problem is playing chess, or writing nonsensical academic crap, or identifying which photograph contains a tank–but is then solved by constructing a machine which clearly does not emulate human intelligence, showing that it was not actually AI-complete. Two things then occur.
On the one hand, some folks (incorrectly) reason that since an AI-complete problem was solved, the robots will be taking over next week. But because the problem was evidently not AI-complete, this conclusion is simple kneejerk ignorance.
On the other hand, simultaneously but in opposition to the first reaction, AI critics reason that, since the problem has been peeled off of the field–has been solved without creating a real intelligence–AI has failed yet again. This is more accurate, but it paints a picture of AI studies as being nothing but a litany of failure. It’s not; the fact is that plenty of AI results have been successful; they’ve just moved the problem domain somewhere it could be solved.
I’m a sysadmin. My job is pretty much to automate everything I possibly can. No matter how complicated the mechanisms I put in place are, they’re simply extensions of the boring parts of my will. Exporting cleverness is AI-complete.
(And as a footnote, I love that writing actual prose appears to be AI-complete, but writing pseudoacademic bullshit is not. Take that, obscurantism!)
GuyFromOhio
It’s just more automation of rote labor. See also manufacturing, automobile. Daleks, terminators, the Matrix, Colossus: how different from bayonets, muskets, the Colt, daisy cutters, tactical thermonuclear devices?
"That’s it. You hit it. That’s control, isn’t it? If we wanted we could smash them to bits … "
Nice to see so many like-minded folk in the wee hours.
Jill
Besides, he’s named after an old Coleco cassette-driven computer from 1984.
Sirkowski
Adam? Jesus Christ, it’s Neon Genesis Evangelion!
Dr. Gaius Baltar
These are splendid developments.
Dr. Frankenstein
mankind never learns
Dr.BDH
What wingnuts wand to know is, does ADAM believe in global warming?
Dr.BDH
"want," dammit.
Atreju
When the robots came for the auto workers I said nothing, because I was not an auto worker. When the software came for the middle managers I said nothing because I was not a middle manager. When the robots came for the scientists…