ZuckerBorg assimilates Microsoft boffins into potentially world-threatening FART

Human-level intelligence?

That sounds great, maybe they can apply their technology to the average Facebook user!

Of course, if they were to make their users intelligent, then they would stop being free content generators and would go out and get a life. That would be bad for the bottom line, so I'm not sure they've really thought this through all the way.

AI users?

OK, I know the growth of Facebook has stagnated, but to use AIs to prop it up smacks of desperation.

On the plus side, maybe AIs will refrain from causing global devastation if they doesn't get enough likes.

SkyNet just started spying. NSA, GCHQ and BND like this.

MS guy that works in AI?

Good lord! They're making Clippy robots!

"Have you noticed how lately ALL humans are starting to look like a threat; not just the ones on the other side?"

178994,4533556,543355 likes (in the last microsecond)

I guess it is natural to over estimate the computational power of the human brain. Doing so leads one to increasingly desperate escalations in the mathematics and hardware used. My view is that a current PC is sufficently powerful to do something like human level AI, the main contraint being the amount of SDRAM available. There are arguments about the 100 billion neurons in the human brain by 1000 dendries making an astronomical computing device. I don't buy it. Likely those 1000 dendrites are used for an inefficient form of random projection. The neurons are likely connected in groups that switch between discrete states (another inefficiency). I doubt the whole thing amounts to more than 100 Gigabytes of in-memory computing. It would also be easier to evolve a system that involved memory look-up using random projects than almost anything else. That would be much simpler to evolve than even a straight forward look up table (LUT) as used on a digital computer.

There hasn't been enough time for evolution to find anything too involved with regard to intelligence in the higher animals. 4 million years by a population of less than a million for most of that time.

I think the problem (if you really want to solve it, and you may not want to) is best handed over to some clearheaded engineers who have not been contaminated too much by the AI research literature. Anyway Facebook, Google and the others are just buying up alchemists and voodoo guys at the moment, which you can say is good news.

"FART"? really?

I understand that it is supposed to mean AI research team, but "FART"? really?

you just cannot take them seriously if the acronym is "FART", I'm sorry but it's not possible to take them seriously

Happy weekend

- cap44

Thought Zuckerberg should have been happy...

... of having made so much money just on Natural Stupidity.

But probably now he needs machines to herd his sheep and extract even more money.

Risks of using Ai to make decisions humans should be making....

* Recently the wife got her credit hit by a fraudulent 10k charge. Three credit cards were issued in her name at a bank that she had no pre-existing relationship with, and therefore no salary or other collateral at.

* Turned out the bank used an automated system to offer pre-approved credit cards to fraudsters. This is great example of automated-system-failure. On the flip side, I can already see examples of couples being denied credit by blunt decision making systems.

* Using AI makes life easy for corporations, but what happens when that same automated system is used to deny or allow health insurance payments for urgently needed drugs or surgery? I see that as a real danger long before any AI mass-killing machines...