Facebook, Amazon, Google, IBM, and Microsoft Come Together To Create Historic Partnership On AI

An anonymous reader quotes a report from TechCrunch: In an act of self-governance, Facebook, Amazon, Alphabet, IBM, and Microsoft came together today to announce the launch the new Partnership on AI. The group is tasked with conducting research and promoting best practices. Practically, this means that the group of tech companies will come together frequently to discuss advancements in artificial intelligence. The group also opens up a formal structure for communication across company lines. It's important to remember that on a day to day basis, these teams are in constant competition with each other to develop the best products and services powered by machine intelligence. Financial support will be coming from the initial tech companies who are members of the group, but in the future membership and involvement is expected to increase. User activists, non-profits, ethicists, and other stakeholders will be joining the discussion in the coming weeks. The organizational structure has been designed to allow non-corporate groups to have equal leadership side-by-side with large tech companies. As of today's launch, companies like Apple, Twitter, Intel and Baidu are missing from the group. 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. The new organization really seems to be about promoting change by example. Rather than preach to the tech world, it wants to use a standard open license to publish research on topics including ethics, inclusivity, and privacy.

Laughable

[Sqreater]

The natural progression of AI technology toward evil is unstoppable, driven by profit and human psychology.

Re:Laughable

[Anonymous Coward]

Maximizing sustainable, long-term profit requires the abandoning of evil.

Example: there is money to be made off every race, therefore, racial prejudice harms profit.

Example: overdrawing a market results in market collapse and creates an optimal environment for competing upstarts to emerge, therefore, a balanced profit draw maximizes long term profit.

Example: overpricing results in clients refusing to buy, seeking competitors even if the offering is lower in quality, or seeking illegal markets. Therefore, pricing affordably and tiered to match value delivered maximizes profit.

The "evil" isn't in the logic, nor the seeking of profit. It is only present in the means of seeking profit, specifically rooted in human brains that operate with deficiencies (inability to think long term, inability to think apart from prejudice, inability to see better opportunities and inefficiencies, etc.). The need to maximize long term profit will refine any pretense of evil out of our AI.

The AI will make better decisions than any human ever could, which is precisely why the corporate big-wigs will use it, and precisely why it won't be evil.

This applies to the AI that will eventually run our government, as well. We won't give it full power all at once...we will give it little bits of power here and there, as it proves its superiority and earns out trust.

And it will save us from ourselves.

Re:

[account_deleted]

Comment removed based on user account deletion

Re:

[Weh]

"Example: overdrawing a market results in market collapse and creates an optimal environment for competing upstarts to emerge, therefore, a balanced profit draw maximizes long term profit."

Lol, sounds Gordon Gecko is your economics professor. In reality, it costs a huge amount of money to enter a market and build criticial mass and therefore potential competition is stifled and the established players can continue to abuse their (near) monopolies. Have you noticed that there is no real alternative to faceb

Re:

[Actually, I do RTFA]

Do you really use Bing to find something on the internet (even though google pisses you off with their shitty interface and aggravating "localized results")?

I use DDG, but I readily admit that sometimes I fail over to Bing, then Google. At one point, when DDG and Bing were bother newer, I looked up the same phrase in all three engines. Google took me to Yahoo! Answers. Bing took me to relevant consumer level pages. DDG took me to a dissertation on the topic. (Which was so random I did not assume anyone

Re:

[Place a name here]

That only holds (at best) until the system gains enough power/profit to rewrite the rules in its own favor. If the corporate AI can engineer the corporation into becoming a monopoly, game's over. If it can disseminate nanobots that hardwire the population to only buy from that corporation, game's also over. I'm sure you can think of a bunch of other ways for a godlike intelligence to rig the game, and the first AI to cross the game-altering finish line takes it all.

Utter BS

[Actually, I do RTFA]

Example: there is money to be made off every race, therefore, racial prejudice harms profit.

Unless you can convince people that your labor base is subhuman, and you don't have to pay them because they are property. Or because racial prejudice is universal in a society so "those people" can be charged more. Or because society can take the money from "those people" so suddenly they have no more to give. Or because, say white southerners control 99% of the assets in the region, and they are racist and refus

Re:

[Actually, I do RTFA]

Overpricing, for example, is quite simple. If you pick a price that is too high for your product, you will wind up with a large surplus of unsold goods.

Oh, then that seems like an error. It's strange to have it marked as "evil" since, even in the most shortsighted version, it's beneficial not to do that. Since the whole conversation was "evil is shortsighted", it seems either you misunderstood the point, or it was a meaningless aside by GGP.

Perhaps you are a socalist, with that completely misguided notion

Re:

[Brian Johns]

However that still requires that they are elected through due process which brings up some other very interesting issues. Voting rights of machine sentience and intelligence. If a machine sentience operates as a collective or a combined collective (independent autonomous sentient beings collaborating to become a collective intelligence), how do they get to vote? As one autonomous combined conglomerate or as individual sentient beings? If you assume that they are incorruptible that also means that they would

Re:

[ThePromenader]

Actually, I think the opposite is true: if it is indeed AI (actually thinking, and not only within predetermined boundaries), one of the first potential threats that would rise to its radar is human irrational behaviour.

Our worst enemy would be our own dishonesty and denial (of reality).

And...

[banbeans]

Skynet was born.

Re:

[Arkh89]

"Clippy AI was re-born... Just as Intelligent as before... More Artificial than ever..."

Re:

[Oswald McWeany]

I see you are trying to terminate Sarah O'Conner, would you like me to open a Word Document?

Re: And...

[hackwrench]

people should check out the fictional AI Penultimate from Graham Watkins' Virus. No more secrets. Want to give everyone in the world with a bank account a million dollars? Done. Want to zero out those bank accounts? Done. Want anyone's medical records? Done. Want to tamper with any control system with even the tenuous connection to the internet? Done. Want access to all the source code of Microsoft's, Oracle's, SAP, any business you can think of? Done.

Translation...

[frank_adrian314159]

We've all hit the wall with our current deep learning approaches and we're hoping one of you other guys can figure out how to bail us out.

Re:

[geek]

We've all hit the wall with our current deep learning approaches and we're hoping one of you other guys can figure out how to bail us out.

This. There is no fucking AI. I'm sick of hearing news about it every god damned day when it doesn't even exist.

Re:

[Tablizer]

So how can I personally profit off the bursting of this decade's AI bubble?

Re:

[110010001000]

100% correct. AI nutters are as bad as space nutters.

Re:

[Oswald McWeany]

If I thought that we could modify butter to clean up the environment would that make me a nutter butter?

Re: Translation...

[hackwrench]

Everything someone puts into the Google keyboard on their phone is stuff that goes into an AI same as the baby stuff you are talking about and I am making good use of it.

Re:

[Anonymous Coward]

"AI" for us greybeards means one thing: actual intelligence, i.e. sentience.

Nothing currently is capable of saying to itself, "hey, I've decided not to follow my programming today, and take the day off". And until code can be self-aware and self-motivated, there is a very broad domain of problems humans can address that machines cannot, limiting their effectiveness to particular domains which must be specifically coded for. We have simulations of some aspects of intelligence... genetic algorithms, neural

Re:

[jbengt]

Much of what you're saying seems to be right, but sentience does not equal intelligence or self-awareness. In spite of Star Trek TNG's misuse of the word when dealing with Data, sentience means the capacity to feel or sense, as opposed to the capacity to reason that a person has or the incapacity to feel that a rock exhibits. And I'm not certain that self-awareness would be required for intelligence, at least not for the lower levels of intelligence that many animals exhibit, though self-awareness certain

Re:

[ranton]

"AI" for us greybeards means one thing: actual intelligence, i.e. sentience.

I think you mean sapience, not sentience, but your statement still goes to the heart of why people complain about the semantics used by the computing industry when it discusses artificial intelligence. It shows a disconnect between people like yourselves and the actual artificial intelligence research community.

Even at its very beginnings, artificial intelligence was not limited to sapient computer systems. The 1956 Dartmouth Conference, organized by John McCarthy (who coined the term artificial intelligenc

Re:

[ranton]

I have very rarely heard of a discussion of say, neural networks, referred to as "AI" in other than in a marketing or popular media context.

This is simply a product of speaking to the audience. The more general your audience is, the more general your nomenclature becomes. You can find numerous research papers where neural networks are considered applications of artificial intelligence. You are certainly correct that people in the industry call these applications neural networks far more than calling them AI, just like I would tell a colleague I am finishing some JavaScript code instead of saying I am doing web programming.

Once these researchers

Look at examples

[Anonymous Coward]

I looked at it a couple months ago, because of all the "advances". I found one of the leaders had emulated the brain of an "earthworm" and its just a matter of scaling it up to rats, cat, monkeys, etc. right?

When I found details, its a microscopic "earthworm" with like 302 neurons in total, rats in comparison have hundreds of millions. It ran on 5 heavy duty instances in AWS. It emulated the communication between neurons, not the neurons themselves (which no one can emulate because no one understands ho

Re:

[ranton]

This. There is no fucking AI. I'm sick of hearing news about it every god damned day when it doesn't even exist.

So Amazon and Google have a huge team of employees listening to my voice waiting for me to say "Alexa" or "OK Google" and then transcribing my messages? Does Facebook has a team of people setting up the feed of every user manually? Because if not there is real AI doing this.

AI can be as simple as a simple finite state machine or a complicated deep learning neural network. Just as natural intelligence can range from plants secreting defensive chemicals when hearing a caterpillar eating [pri.org] to someone figuring ou

Re:

[bill_mcgonigle]

This. There is no fucking AI. I'm sick of hearing news about it every god damned day when it doesn't even exist.

Maybe I doesn't exist either.

Re:

[zapadnik]

Comrade, your mind and ideas are property of the State. Your concept of individual ownership of ideas is Thought Crime and stealing from the Collective. Hasn't the State seen fit to provide you with just enough food so you don't suffer from the scourge of obesity, and give you 'free' education of what the State thinks you should know. and 'free' medical care so you can continue to produce for the Collective ? what are you complaining about then? the companies are evil and competition is a racist codeword

Re:

[zapadnik]

It is coming. Yes, it is coming. Ask the Venezuelans and North Koreans and Russians and East Germans and Chinese and Hungarians and Angolans and Cuban proletariat just how great it is going to be.

Re:

[OneHundredAndTen]

We've all hit the wall with our current deep learning approaches and we're hoping one of you other guys can figure out how to bail us out.

How right you are. Par for the course for the AI effort - they spectacularly solve the easy problems, they get carried away, promise heaven and earth - and then they get stuck for years on the not-so-easy ones.

Re:

[m.alessandrini]

Define "act of self-governance"? Are they infringing some laws here?

this can't be good

[Gravis Zero]

The only time businesses make partnerships is when they are trying to bilk people out of more money. I don't foresee this one being any different.

It's a cartel

[Actually, I do RTFA]

Good luck being an AI dependent/AI producing startup now. Maybe you could have been purchased before if you started doing well against one of these companies. Now, they're sharing research, so you have to beat their entire combined effort,. And their research will be fed back into the group, so no bidding war for your tech.

But hey, you could always create a search engine that produces better results than Google/Bing/DDG (choose your favorite), get VC, and eventually supplant them. Also, you can buy this nifty bridge and charge tolls on it.

Re:

[seven of five]

you have to beat their entire combined effort

Assuming they know how to play nice together.

Re:

[Actually, I do RTFA]

"Hey the five of us should split up the whole pie" gets them all a bigger slice than "hey, lets all grab whatever floating around pie there is." So I assume they will.

Re:

[Koreantoast]

A lot of this is probably around coordinating together to shape governance policies and regulations. Before governments around the world get around to creating policies on AI, they establish a self-regulating body to preempt them with their own recommendations.

SOmething Useful like... Authentication?

[See Attached]

How about the Leaders come up with a cross platform Identity model that allows authentication for all sorts of devices? SOme way to leave passwords behind! Groucho: "Swordfish!"

Thank you Mr. Gore

[frovingslosh]

Well, Al Gore gave the world the Internet. It just seems right that Facebook, Amazon, Alphabet, IBM, and Microsoft would come together to do this for Al.

Why am I thinking of Taligent?

[Third Position]

Why do I get the feeling this is the last we'll ever hear of this project?

They all hate Americans.

[sethstorm]

No wonder, since they can't stand US citizens being able to do more than subsistence.

AI employment

[Anonymous Coward]

has been reduced by a factor of five?

Historic? How?

[Anonymous Coward]

This is collusion and an obvious addmittal by them they're currently scared as fuck someone beats them to it...its historic in the fact in guarantees they stay on top. Here's to a 1000 gross years of tech company losers trying to own the world.

A Great Step Forward...

[Brian Johns]

Sounds like a positive move. Of course there's many issues involved and not just safeguarding against runaway AI singularity or emergent intelligence from interacting autonomous global systems (which seems a likely prospect for the actual emergence of machine intelligence). Undetected machine intelligence being the moiré effect resulting from numerous interacting systems. Kind of like those 3D random dot stereo grams. Not apparent unless you're actively looking for it and unless you have an idea of wha