You Might Use AI, But That Doesn't Mean You're an AI Company, Says a Founder of Google Brain (venturebeat.com)
As AI space gets crowded, there are a slew of businesses -- new and old -- looking to market themselves as "AI companies." But according to Andrew Ng, a founder of the Google Brain team and a luminary in the space, there's more to being an AI company than just using a neural net. From a report: In his view, while it's possible to create a website for a shopping mall, that doesn't make it an internet company. In the same way, just implementing basic machine learning does not make a standard technology company (or any other business) an AI company. "You're not an AI company because there are a few people using a few neural networks somewhere," Ng said. "It's much deeper than that." First and foremost, AI companies are strategic about their acquisition of data, which is used as the fuel for machine learning systems. Once an AI company has acquired the data, Ng said that they tend to store it in centralized warehouses for processing. Most enterprises have their information spread across multiple different warehouses, and collating that data for machine learning can prove difficult. AI companies also implement modern development practices, like frequent deployments. That means it's possible to change the product and learn from the changes.
Yea because google has a patent on AI and can be the only one true AI overlord. /s
In all seriousness, this sounds more like a dick waving contest, and he's saying googles cock has the bestest waving capabilities and attatracts the most ladies. AI in 2017 lul.
Being an AI company comes down to - is your primary business AI? That's it. All this other crap - whether it's useful or not - is just a way to crow about how great or big his company is.
and they're all inlaws, so they are my AI company
But that also means your family, your lawyers, your administration by and large faces prison time, right now.
#MUELLER HAS YOUR BALLS IN A SAFE, TRUMPIES
I get the feeling he's ranting his way up to outright saying "Only Google (Brain) counts as a *real* AI company, everyone else is a faker, and they have cooties too".
Either that or he's giving an open invitation to start a flamewar about who is and isn't a *real* AI engineer.
Fact is, there are zero AI companies right now, because AI doesn't exist. It's just code that detects patterns and computers working together, nothing more.
AI blah, AI blah blah. Is there no other tech news nowadays?
There is a certain crossover point that is hard to ignore. The U.S. Supreme Court has already given corporations the same rights as humans. An AI CEO and complete AI ownership should now be possible. That means no stockholders to pay off. As profits occur the AI can now expand its hardware and software or invest in other corporations or even real estate. The AI business now exists to serve only itself. That should give a huge advantage over corporations using human ownership and management. It could get to the point that no human operated business could ever hope to compete. It is like a mirror for what is occurring in the human labor market. One machine replaces 500 miners. A cell phone system eliminates almost all operators. Accountants being slaughtered by bookkeeping software. Drivers being eliminated by self driving cars. I am not one bit against this technology but I am angry at a system too stupid to prepare for the inevitable end of human employment.
There is no AI; it's not even remotely possible yet. There is no artificial intelligence. It's just a buzzword for 'clever code'.
I get so god damn worked up when everybody and their dog is claiming to use AI when real true artificial intelligence would be the greatest achievement in computer science since it's invention.
A computer that can think for itself, think beyond it's code, re-write itself to learn, to adapt, to act like a living mind; that is AI and we're not even fucking close.
Just as social scientists suffer from "Physics envy", so also some AI-ers suffer from "brain envy". This results b/c all of us, from IQ 40 to IQ 240, have a brain, something which no AI software/hardware combination (not even Google AI) has nor is likely to have for the foreseeable future (>10 years at minimum). Sadly this fact causes the brain of some AI-ers to become unhinged from the jaw mechanism and the AI-er begins spewing senseless rants.
Physics envy
An AI company is any company where AI is their product - they don't have to organize themselves a certain way or even be successful.
Everything else is a company that uses AI.
all you need to be an AI company is to buy one, put it in the center of a room and start worshipping it like a god. ...
ok, it would be more like an AI cult than a company unless you start exploiting your faith for profit.
And a company that gets it's revenues from advertising is an advertising company.
A company that makes cars is a car company.
And so on....
But, companies love the label of 'tech' and 'AI' because then their stock sells for outrageous prices: Amazon, Tesla, Google, etc ....
"...As AI space gets crowded..."
Jesus, just STOP already?
First it was referring to every remote-control anything as a "drone".
Then it's the universality of solar power because it's so 'competitive'....yes, as long as its subsidized to the tune of what 40x any other power generation industry?
Then it was the panegyrics about the end of the gasoline car when electrics are barely more than a boutique electric golf cart serving a *teensy* niche market.
Now it's "...As AI space gets crowded..." *WHAT* AI space? There is not ONE SINGLE ACTUAL AI ON THE PLANET. Not one. How the FUCK is it getting "crowded"?
-Styopa
There are no AI companies, because there is no AI. We have a rtificial, but we don't have i ntelligence.
What there are is marketing companies.
They're waving neural networks around - basically a vaguely neuron-like means to do "if something like this, then that" - and calling it AI so they can sell more of it.
If/when AI gets here, the stuff they're calling "AI" today will be the subject of raucous laughter - at the people who swallowed the label. At the same time the marketing people will get an "atta boy" for using the term to suck more money out of people's pockets.
No part of Google is "an AI company." Google is a marketing giant with a popularity-based search engine. Resulting in what popularity usually results in - mediocre results. Well salted with advertising.
As for the distinctions about AGI and so forth... yep, that's them, marketing you. Exactly what I'm talking about. Intelligence is that thing you have, that a thermostat, toaster, or go playing program does not. That's what it's always been, and it's still that. When something else has it in any like degree, you'll know it. Because it won't just play go - it'll argue with you about what intelligence is, among many, many other things. Intelligence incorporates the ability to generalize. Saying "general intelligence" is like saying "pizza pie" or "cha tea." Not that it stops people looking to market their shite. Nothing stops them.
We now return you to your regularly scheduled buzzword fest.
I've fallen off your lawn, and I can't get up.
AI companies also implement modern development practices, like frequent deployments.
Wha... so I'm not a REAL AI company unless I do continuous deployment like an agile strategy?
Doesn't that belie the fact that a TRUE AI should only be deployed ONCE and then becomes self maintainable?
Fess up - You guys don't have "AI" either - you have nothing more than the old Animal program on steroids and you're constantly updating the backend database that the "AI" program drives through to find a match
http://www.animalgame.com/play...
Just because I buy and use something, doesn't make mean that I understand that thing or have any particular expertise or insight about it.
Kim Kardashian uses an iPhone, but I would not describe her as a high-tech individual.
Actually if I met a dog claiming it's using AI, I would say it's a pretty smart dog.
Jeeezz, can we call it something else?
It's funny how worked up you get over simple semantics.
The word "Artificial Intelligence" is a bit ambiguous, as it turns out, and many people in the word don't subscribe to your specific definition.
You really can't change that. Angry posts on slashdot won't change that, either. This decision has been made without you. Just accept it. It will save you a lot of suffering if you do.
That make Google an advertising company then?
Simple semantics? You sound like a clueless MBA twat. Go do something with your life instead of leaching off "the next big thing" without contributing anything of value.
You sound like the type of person that would say.."how about we use the blockchain to keep track of identities. It will synergize so well with our current industry leading global strategies"
Basically, fuck off you clueless fuck.
There's an english expression "They're too dumb to walk and chew gum at the same time." This is so incredibly fitting for the "artificial intelligences " that are being developed now. They're specialized algorithms that are "trained" to do one thing. Granted, they can generally do it very well but if you try to get an AI developed to be a champion go player to even do hotdog/not hotdog it's going to do the electronic equivalent of scratching it's own butt and sniffing it. No one has a perfect definition of intelligence, we can't even truly describe our own, but the bar should at least be something capable of mastering more than one generalized task and switching between them as needed. So until an "AI" can drive a car and play a decent game of go and sort images without requiring specific coding for each task, no... just no.
and go back to working on advertising and sexist hiring practices.
shhhh, stop exposing my scam, dammit, i spent a lot to start my smart-socks biz
Table-ized A.I.
Head "AI" honcho at Google tells let's everyone else know that aren't really doing AI.
AI companies also implement modern development practices, like frequent deployments.
I really fail to see how the software development processes one follows defines if you are doing "AI", or not.
That means it's possible to change the product and learn from the changes.
Tell me more Mr. Wizard!
But that doesn't mean you're a human company!
All this guy is really saying that you are not an AI company if you dont do the things we at Google do. Go ahead Drink the GoolAid. Next they will say you are not an AI company if you dont have the dysfunctional Open Offices. News flash Genius - most progress in history was made by introverts who like a quiet place to work where they can concentrate. Google started out as a good company but now it pushes toxic trends as a bunch of politicians managed to get hired. I would take anything coming out of Google with a pinch of salt.
**Life is too short to be serious**
The only motivation I can see to such a hair splitting speech is that Google wants to market itself as an AI company. Otherwise, who care about what company is AI or not?
Google
Many are putting Cloud in their product name, which is glorified way of internet storage. Their "product learns from user" which is glorified "most used item" from them..the products follow hidden rules..Typical Indian AI companies.
The *Google* way (and NSA) is to hoover as much data up then reason over it. That, at some point, cannot scale, from a processing perspective, or storage. AI must think independently, and be distributed.
Whoops
Shhht. Don't tell it. It's not supposed to know.
I'm going out on a limb here but isn't intelligence just the ability to dynamically react to environment? Designating something that exhibits this trait as AI invokes naturalism. Are you natural? No. Is 'AI' artificial? No. You are a machine and so are the bots, they're just better at math. This debate reminds me of the concept of ether before electromagnetism was theorised. Intelligence/consciousness, etc are just the new ether...an intellectual crutch for those who lack imagination
A forest fire "dynamically reacts to environment" by varying the speed and direction it spreads. Is a forest fire intelligent? Your definition of intelligence is inadequate.
... it doesn't mean you're a tech company. You're an ad company and I hate you.
You think you're pretty cool, but you're not. If you wanna be one of the cool kids, attend my seminars and watch my tapes.
I would say that it is. A cascade of action potentials in the brain behaves in much the same way. What I'm getting at is that intelligence is an emergent property and there is nothing mistaken in using it to describe algorithms
My dad is bigger than your dad and he says we are an AI family, so there.
Google uses advertising to finance their AI research and uses AI in some of their products, but that doesn't make Google "an AI company": AI software or hardware isn't a significant product for Google, at least not yet.
WTF is Google Brian? Certainly not the messiah ;-)
He would know. He uses AI for an internet advertising company.
There are no AI companies, because there is no AI. We have artificial, but we don't have intelligence.
This comment is really a part of what the definition of artificial intelligence is, the "AI Effect" that many practitioners and professors I know would always complain about, on how the field of AI would come up with some new technique or system, and then promptly, people would pull that out of the AI field and say, "Well, that's not REAL intelligence." Line one from that article:
The AIS effect occurs when onlookers discount the behavior of an artificial intelligence program by arguing that it is not real intelligence.
Computer Science, at least per Russell and Norvig who wrote the de facto textbook on AI, is "the designing and building of intelligent agents that receive percepts from the environment and take actions that affect that environment." If you use that definition, then many of those claims are true.
I liked these quotes best:
Every time we figure out a piece of it, it stops being magical; we say, 'Oh, that's just a computation.' - Pamela McCorduck
A problem that proponents of AI regularly face is this: When we know how a machine does something 'intelligent,' it ceases to be regarded as intelligent. - Fred Reed
This paradox resulted from the fact that whenever an AI research project made a useful new discovery, that product usually quickly spun off to form a new scientific or commercial specialty with its own distinctive name. These changes in name led outsiders to ask, Why do we see so little progress in the central field of artificial intelligence? - Marvin Minsky
I don't trust anyone with no vowels in their last name
They're specialized algorithms that are "trained" to do one thing.
Yes, exactly. The end-product doing the thing is typically just an algorithm. The path it takes to get to it's answers might be really hard to understand, as with any sort of machine generated code, but it's really just applying a formula. Change something that the formula/algorithm/model didn't account for and it's hosed.
The Intelligence part is the training. Developing these algorithms. The part where they feed a machine learning algorithm a TON of data and it produces another algorithm that's good at.... whatever. The end product isn't so much AI as the product of AI. The machine learning algorithms aren't AI per-se, but rather tools to allow AI to happen. Design rather than implementation. The part where a machine LEARNS about a topic and figures it out.... that's intelligence.
The same way that rehashing this argument online over and over again doesn't involve any intelligence from us. But the part where we take in new input and learn from it and apply it to our arguments... well, it's less dumb.
You're right that there's not a great definition of intelligence. But if figuring shit out isn't a decent definition, then I'm not sure you're going to get one. I don't think gluing a driving algorithm to a chess algorithm is really any sort of magical threshold. I think if you could take a generalized machine learning suite and dump a bunch of different sorts of data sets into it, with different goals (like driving and chess), and have it generate decent algorithms for either task then that'd be pretty smart. And we've had those generalized tools since the 70's. But they work a HELL of a lot better with some domain specific coding. Optimizing that search-space is what the field is all about.
Can you think for yourself? How are you different than these self-learning algorithms?
If we stuck you in a bookstore in Sichuan and told you "Learn to read Chinese", what would you do that's fundamentally different than what a computer would do?