4 ms·
I think that the event will help to boost investors confidence, and hence investment in AI research. Which is great, because it means accelerated pace of innova
by nzonbi 11y ago
I think that the event will help to boost investors confidence, and hence investment in AI research. Which is great, because it means accelerated pace of innovation. It is a huge win for the AI field. AlphaGo showed that a special arrangement of neural networks with additional algorithms can deliver spectacular results. They can now try more sophisticated arrangements of neural networks, to achieve more ambitious results. I hope that we will now start to see more creative attempts at AI. The time to finally reach to general AI, will depend on the magnitude of the research effort that is put on. This kind of highly publicized events are very positive.
AlphaGo was a combination of neural networks with a tree search algorithm. I think that very interesting things could be achieved, combining neural networks with basic knowledge representation systems -symbolic AI-. These techniques are highly discredited for under-delivering in the past. But I think that it is a good moment to revisit some of these past techniques, and combine them with the more recent techniques of neural networks. It could be an interesting base to attach neural networks. And then submerge the AI on a basic simulated world. Something like minecraft, perhaps, as Microsoft announced that is going to do. I think that giving the neural networks some basic structure to depend on, can help to achieve results more easily.
In any case, I hope that high profile AI events like AlphaGo victory continue happening, to help to further increase AI research.
- emcq 11y agoThis result came out of a large team that Google paid a lot of money for. Google Brain has been kicking ass internally solving really hard problems like realtime translation or ranking. Most major companies from Toyota to Baidu to IBM have research money focused on AI. Industry and investment is already focused on AI. Relative to other research fields there is not a shortage of investment in AI, and despite the difficult market you see recent big investments in companies like Cruise or OpenAI: https://www.oreilly.com/ideas/the-current-state-of-machine-intelligence-2-0 https://www.oreilly.com/ideas/the-current-state-of-machine-i...
- Ericson2314 11y agoAs a programming language specialist, I am for all things symbolic too. http://leanprover.github.io/presentations/20150717_CICM/#/sec-53 http://leanprover.github.io/presentations/20150717_CICM/#/se... The last slide hints at proof assistent + ML. I wish I knew more of what they are up to!
- amelius 11y agoI'm not sure if I'd like to see AI in a compiler backend; development might become unpredictable (in terms of program efficiency).
- ant6n 11y agoIt kind of already is.
- Ericson2314 11y agoMy guess is the machine learning is some sort of tactic. So it communicates with Lean to generate some code (program/proof), but ultimately Lean itself decides whether the code is correct---this is no "polluted backend".
- Ericson2314 11y agohttps://intelligence.org/2016/03/18/seeking-research-fellows-in-type-theory-and-machine-self-reference/ https://intelligence.org/2016/03/18/seeking-research-fellows... !!!