4 ms·
There are a few things wrong with this sentiment: * The risk of interacting with the first AGI is even harder to put an upper bound on. You might destroy the
by endtime 16y ago
There are a few things wrong with this sentiment:
* The risk of interacting with the first AGI is even harder to put an upper bound on. You might destroy the human race.
* Statistical NLP and ML might be useful "addons" to an AGI but they're unlikely to lead to its creation on their own.
* Google has a lot of really smart engineers and computer scientists, but AGI is neither an engineering nor a CS problem.
- jerf 16y ago"AI is hard" is part of the reason why the profit potential is hard to bound; if it were easy it would also be very easy to assess the profit potential: It would hardly be profitable at all, it would simply be the economic baseline. And I don't see how "it's dangerous" contradicts "it's hard to bound the profit" either.
- endtime 16y agoI don't think I said anywhere that Google shouldn't do it because it's hard. I said they shouldn't do it because it's dangerous and because it's not the kind of problem they know how to solve.