Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Facebook, Amazon, Google, IBM and Microsoft Create Partnership on AI

techcrunch.com

Post reply on HN
Print view View original

241–250 of 270 posts

Re: Facebook, Amazon, Google, IBM and Microsoft Create Partnership on AI

#241

Post by ptero »

wu-ikkyu wrote: post #229
ptero wrote: post #207

Earlier quoted context omitted.

They chose their initial partners somehow and did not include OpenAI. Maybe they thought this would make everything too transparent, maybe they see little financial benefits (those are private companies in general aiming to benefit shareholders), whatever. I see no problem with this. I wish OpenAI plays there eventually, but even now some of the partners go beyond what I would expect from a private company on fosteri…

Why does it matter whether it is a government agency that has those data sets versus a private corporation?

Because: (1) government can ignore some protections that corporations are supposed to follow and (2) government can act on this in a way corporations cannot (use this information to argue for or issue search / seize warrants, etc.).

Re: Facebook, Amazon, Google, IBM and Microsoft Create Partnership on AI

#242

Post by paisible »

ionwake wrote: post #178

Which is the best way to invest in the field of AI ? Can anyone recommend any specific companies / branches ? Thank you.

One way is through hardware that supports the current wave of innovation, i.e : GPUs.

Nvidia has been on a tear recently. I'd love others' input on this.

Re: Facebook, Amazon, Google, IBM and Microsoft Create Partnership on AI

#243

Post by wu-ikkyu »

ptero wrote: post #241
wu-ikkyu wrote: post #229

Earlier quoted context omitted.

Why does it matter whether it is a government agency that has those data sets versus a private corporation?

Because: (1) government can ignore some protections that corporations are supposed to follow and (2) government can act on this in a way corporations cannot (use this information to argue for or issue search / seize warrants, etc.).

>government can ignore some protections that corporations are supposed to follow

Vice versa is also true. Google and Facebook can censor free speech more effectively than any other, legally.

>government can act on this in a way corporations cannot (use this information to argue for or issue search / seize warrants, etc.)

Google and Facebook don't need a search warrant to scour and social engineer billions of people's lives.

Re: Facebook, Amazon, Google, IBM and Microsoft Create Partnership on AI

#244

Post by eridius »

wstrange wrote: post #110
eridius wrote: post #72

Earlier quoted context omitted.

At WWDC they talked about how they put AI all throughout the system. It's not just when talking to Siri. It's things like suggested apps, photo tagging, deciding when to refresh apps in the background, text prediction, and probably lots of other stuff that I can't even think of because Apple's approach to AI is to make AI largely an implementation detail of the experience rather than a user-visible component.

Apple makes great stuff (typing this on a Mac), but I really don't get the adoration / mystic that surrounds them. Those features have been on Android forever, but people don't call it "AI".

In this context, "AI" means "using techniques that come from the field of AI research", such as convolutional neural networks and whatnot.

Re: Facebook, Amazon, Google, IBM and Microsoft Create Partnership on AI

#245

Post by Taek »

weaksauce wrote: post #238
Taek wrote: post #164

Earlier quoted context omitted.

Modern machine learning depends heavily on data. If you have 100x the data, your algorithm is going to outperform a superior algorithm in most cases, because the data is the source of the intelligence. It's a big, scary barrier for startups, for innovation, and we're running the risk of creating data superpowers that can't be competed with. Some algorithms don't perform well until they have a threshold of data. When…

> But we do know that they are using their position to manipulate ISIS recruitment, which means they aren't afraid to get their hands dirty with this kind of stuff To be fair, Google is not cooking the search results, a company under their wing(jigsaw, formerly ideas) is buying ads on search terms related to terrorism. How good a buy deal they get is one thing to consider and I heard there was a grant of Adwords mone…

Might be controversial, but I pretty strongly believe that buying ads is in the same category of manipulation as altering search results, with perhaps the caveat that ads which declare themselves as ads are maybe not nearly as bad.

Ads manipulate the flow of information going to a person for the benefit of the advertiser. It's taking advantage of a vulnerable person in hopes that they will change their behavior - whether that change in behavior is beneficial is irrelevant.

Re: Facebook, Amazon, Google, IBM and Microsoft Create Partnership on AI

#246

Post by Taek »

parasubvert wrote: post #194
Taek wrote: post #164

Earlier quoted context omitted.

Modern machine learning depends heavily on data. If you have 100x the data, your algorithm is going to outperform a superior algorithm in most cases, because the data is the source of the intelligence. It's a big, scary barrier for startups, for innovation, and we're running the risk of creating data superpowers that can't be competed with. Some algorithms don't perform well until they have a threshold of data. When…

"THERE IS NO OVERSIGHT!" Should there be? I'm not quite sure. I mean, personally I feel a lot safer managing my own data privacy with what I give these companies, rather than having a government do it for me. And I'm not exactly a libertarian, I just think it's rational that this should be decentralized, the risk of a company turning against "us" is a lot less than a government, given history. "I've switched away fro…

I don't think you need oversight if you have a fair playing field for competitors. If switching from Facebook to... that other social network (?.. that's exactly the problem) was as easy as switching which restaurant you eat at, I wouldn't be calling for regulation I would be content using the social network that I thought was good for me. But I can't, because nothing comes close to Facebook.

I feel that if we had broken up Microsoft in the 90's, we'd be in a much better place today. We're still churning out professionals by the millions who are fully dependent on proprietary software to make a living, when nearly equivalent free alternatives do exist for many of them (Office, CAD, Adobe are the biggest). It makes me sad, and Microsoft had a huge hand in that.

Re: Facebook, Amazon, Google, IBM and Microsoft Create Partnership on AI

#247

Post by gsmiro »

parasubvert wrote: post #194
Taek wrote: post #164

Earlier quoted context omitted.

Modern machine learning depends heavily on data. If you have 100x the data, your algorithm is going to outperform a superior algorithm in most cases, because the data is the source of the intelligence. It's a big, scary barrier for startups, for innovation, and we're running the risk of creating data superpowers that can't be competed with. Some algorithms don't perform well until they have a threshold of data. When…

"THERE IS NO OVERSIGHT!" Should there be? I'm not quite sure. I mean, personally I feel a lot safer managing my own data privacy with what I give these companies, rather than having a government do it for me. And I'm not exactly a libertarian, I just think it's rational that this should be decentralized, the risk of a company turning against "us" is a lot less than a government, given history. "I've switched away fro…

For an example of how oversight do nothing for the end user take a look at the Brazilian model of state regulation "agencies" and how they get influenced by the bigger companies.

Re: Facebook, Amazon, Google, IBM and Microsoft Create Partnership on AI

#248

Post by phodo »

modeless wrote: post #23
phodo wrote: post #5

At the risk of ad hominem, this is typical techcrunch reporting: >> "Though Apple is said to be enthusiastic about the project, their absence is still notable because the company has fallen behind in artificial intelligence when compared to its rivals — many of whom are part of this new group." How exactly is it that TC knows that Apple has indeed fallen behind? Are they privy to the Apple ML roadmap? Are they using…

> How exactly is it that TC knows that Apple has indeed fallen behind? TC is not alone in saying this. It's a very common sentiment in the machine learning community. So much so that Apple recently did a big PR piece in Backchannel just to try to refute it. https://backchannel.com/an-exclusive-look-at-how-ai-and-mach... > there were more or less objective (err, almost objective) benchmarks [...] This is not the case…

To clarify, this was not in reference to the lack of metrics in the field, which exist, obviously. Nor was it in reference to the lack of metrics / citations in academic and scientific publications. This was in reference to a lack of specific citations/metric backing up a claim, from a popular publication targeting a mass audience does not necessarily have the subject matter expertise to fill in the data on their own. The information you provide comparing Apple's lack of benchmarking to others would have been useful to have in the article.

Re: Facebook, Amazon, Google, IBM and Microsoft Create Partnership on AI

#249

Post by phodo »

aub3bhat wrote: post #50
aleeds wrote: post #16

Earlier quoted context omitted.

"But in Machine Translation, Apple is ahead." was an example of a statement that could be made if metrics existed, he is not asserting that that is true.

I assumed that he was asserting it to be true, because as my updated comments shows, the metrics do exist. But Apple neither releases them nor does it discloses the architecture/technology (if any) used.

As others have mentioned, you are completely wrong in your assumptions. It is you who shows zero clue in reading comprehension and communication. I'm well aware of the benchmarks used by industry and academia. I've been working in the industry for many years, including multiple ML-based products used commercially and running in production serving a large number of customers. Machine translation was just an example, and there was no assertion that company A is better than B; I was in no way "asserting it to be true". I was merely providing an illustrative example to highlight the fact that * popular press * articles about ml, such as this one, are quick with subjective opinion, without providing in writing the necessary citation and metrics to back their claims. Furthermore, they do so at a generalized level, without going into specific subdomains. My initial post was intended as a constructive comment, so take it as such.
Post reply on HN