7 ms·
Inside Darpa’s Push to Make Artificial Intelligence Explain Itself
- randomerr 9y agoIf DARPA can't it's own UI, what good is it?
- arcanus 9y agoThe government is far behind on ML/AI. I don't see any evidence this effort is going to help them catch up. In fact, the department of defense is far behind even the department of energy inside the US government on high performance computing, algorithms and machine learning.
- Sangermaine 9y agoHow can you know this? Would DARPA or the DoD tell us about their advances?
- castis 9y agoI would imagine that the publics knowledge about their advancements is significantly far behind what it actually is.
- SomeStupidPoint 9y agoIsn't this fairly normal? I seem to recall that the DoE, NOAA, and NASA always had a lot of computing resources, and that they generally rented time on those clusters out to other government agencies. I do wish the government would step up on AI though -- I think that they owe it to the country to put together a Manhattan Project for AI.
- arcanus 9y agoPerhaps normal, but the concern is that this could be a strategic disadvantage versus countries that are massively boosting these programs. The Chinese government is leading (and funding) massive efforts in high performance computing and AI. If they win, entire high tech industries could be situated in (and lead by) China.
- SomeStupidPoint 9y agoI'm troubled by the DoD actions on AI, but not by who owns the supercomputers in the US. It actually works out better if the US government computers are through various science organization labs, because it locates much of their computing power in places also accessible to universities and at agencies amenable to US universities using their resources. So funding computers through the DoE, NOAA, NASA, etc actually makes them more available for the US research community, while allowing the military to lease the time they need (eg, for calculating explosion dynamics). The total funding for US supercomputers also could be higher, but that's a general problem with US infrastructure investment. My main point was there's no real concern with the computers being owned by scientific institutions instead of the military -- they're still government funded, it just makes them more widely accessible for use.
- wu-ikkyu 9y ago>I think that they owe it to the country to put together a Manhattan Project for AI. I hope we can come up with a better primary goal for AI than mass murder
- BatFastard 9y agoI first learned about Machine learning from Darpa in 2008, so I doubt they are behind
- lolokaydude 9y agouh, machine learning has been a going thing for decades. Ray Kurzweil was selling OCR in the late 70s.
- BatFastard 9y agoWell it was called AI back then.
- pedrosorio 9y agoNot even close. Also, machine learning is a subfield of AI. https://cs.stackexchange.com/questions/37618/who-coined-the-term-machine-learning https://cs.stackexchange.com/questions/37618/who-coined-the-...
- pizza 9y agoI highly doubt the government (or, at least DARPA,) is. See e.g. https://www.darpa.mil/attachments/AIFull.pdf https://www.darpa.mil/attachments/AIFull.pdf, or https://www.darpa.mil/program/probabilistic-programming-for-advancing-machine-Learning https://www.darpa.mil/program/probabilistic-programming-for-...
- lolokaydude 9y agoNeither of that are impressive? The issue is that they don't publish a ton of papers, but TBH those look like the projects of an organization that is slow moving and behind the times.
- mattkrause 9y agoYou are not going to see many research papers with DARPA staff as an authors. That's just not how they work. Instead, they contract with academic and industrial researchers to work on specific things that are (or might someday be) of particular interest to the DoD. Thus, if there's a huge DARPA-funded breakthrough in Probabilistic Programming, it's going to have (say) Frank Woods' and Oxford University's name on it, not the DARPA program manager's. The final work might not even be published. They're happy to fund industrial groups if that lays the groundwork for products that the DoD eventually wants to buy.
- pizza 9y agoThey're also (currently?) running a competition for distribution of radio bandwidth amongst peers using ML, iirc
- nl 9y agoDARPA funds work, it doesn't do it itself.
- mattkrause 9y agoHow would you know? I would guess that the DoD is not particularly interested in releasing lots of information about their capabilities. Second, the Department of Energy isn't just a random department with a lot of spare computing power. They run the National Labs, and part of that work is to make supercomputing available to academic and industrial researchers too. These are also used for nuclear-related projects (part of the DoE's mandate, but obviously related to DoD concerns too) and I would bet that there is some mechanism to boot off academic researchers if the national interest suddenly required exaflops of computation,
- fapjacks 9y agoThis is actually something pretty easy to learn. If you pay attention to the business solicitations on FedBizOpps [0], you can see the whole swath of what each DoD entity wants to achieve in the next year or two, because they lay it all out explicitly. [0] https://www.fbo.gov/ https://www.fbo.gov/
- torbjorn 9y agothis is a fool's errand. today's statistical inference based "AI" is a probability distribution that squishes itself through cracks in data to "learn" a path of least resistance. subject-object reasoning systems, such as English, are cut from a different cloth. semantic language is a discontinuous function. Asking asking an entity that can only discern its environment through continuous functions to explain itself is like asking an amoeba to play piano. Machine Learning is a statistics renaissance, it is changing the world, but "Explainable AI" is a notion that is totally divorced from where the field is at atm.
- FullMtlAlcoholc 9y agoOr, it's like expecting a mathematical function to solve itself
- dsjoerg 9y agoIt's easy to be dismissive, but the project becomes more tractable when you think of the task of "explaining" as yet another classification task. The goal is not necessarily to explain _exactly_ what's happening inside the AI, no more than a human can explain their decisions by describing _exactly_ what's happening inside their heads. The explanation is an additional _decision_ that accompanies the original classification decision. The explanation space has its own utility function and the goal is to find the explanation with the highest utility.
- torbjorn 9y agoI don't mean to come off as dismissive, and yes my comment is a sassy one. Machine Vision systems are proof that neural networks and other machine learning techniques are revolutionary. I really believe that. But the notion of "Explainable AI" just reeks of upper middle management folks performing what they see as visionary risk management. When in reality these people don't know what they are talking about. Explaining exactly what's happening inside "the AI" is the domain of linear algebra. We can't be as exact in plain english but the generalization "it's fitting the training data" speaks volumes more than any _just_around_the_corner_ "explainable AI" secret sauce algorithm ever will.
- giardini 9y agoThe URL is behind a subscribe wall. Is there an open link available?
- danschumann 9y agoAs long as it doesn't realize it's easier to lie to humans about what it's ACTUALLY doing. Maybe make the explainer algorithm is non-ai-based, to ensure it doesn't optimize out honesty.
- Retra 9y agoCouldn't you just compartmentalize? Make an outer AI that has to defer to an inner AI, and nest them in such a way that the only interface between the layers is a 'explaining' communication channel?
- therein 9y agoThat reminds me of this video and the left/right brain compartmentalization. https://www.youtube.com/watch?v=wfYbgdo8e-8 https://www.youtube.com/watch?v=wfYbgdo8e-8
- nradov 9y agoIt seems a bit unfair to hold AI to higher explanatory standards than our fellow humans. Most people make decisions based on intuition and emotion. Then they rationalize those decisions after the fact and invent explanations that seem plausible.
- nl 9y agoThis is unnecessary anthropomorphizing. Explaining how an AI system makes decisions is completely unrelated to any kind of progress (or not!) towards AGI.
- roceasta 9y agoHaven't read the article but what if General Intelligence precisely is the ability to explain oneself to oneself (or to others)? This could turn out to be a chicken and egg problem.
- major505 9y agoDo you wanna the Judment day? Because that's how you got a Judment Day.