Where is the artificial intelligence (AI)?

[DTEJD1997]

Hey all:

 

Similar to my thread on UFO's, where is the artificial intelligence?

 

In my day to day life and activities, I see almost no hint of it.

 

For example, my credit cards is frequently declined, even though I have TONS of capacity on it.  Turns out it is a "safety" thing.  The crazy thing is that if whatever program/algorithm is running this had even a modicum of "intelligence", it would be IMMEDIATELY apparent that it is me, and a "normal" level of activity.  For example, I will visit merchants across town, at roughly the same time of day, and in the same order.  I have done this NUMEROUS times over the years...any human, with any level of comprehension could see the pattern and understand that I am guying things for my business...but the algorithm running the fraud detection throws a fit.

 

Conversely, Ebay & PayPal have had MASSIVE breaches of their accounts/databases in the past couple of months.  It seems to be dying down now, but at some points I was having more fraudulent sales than legitimate sales.

 

The real problem was that it was STUPIDLY OBVIOUS that the sales were fraudulent (well 95%+ of the time).  The fraudulent transactions had many similarities that even a 1st grade human could easily see & detect.  The most glaring of which was that the scammers were requesting delivery to the same address!  To further add to the idiocy, this address is well known to the interwebs as being a nest of thieves.  If you were to Google that address, the first page of search results comes back loaded with sellers on Ebay, Amazon, and others sites complaining of a similar thing to what I experienced.  If you look at the times stamps for some of those websites, some of them are a year+ old!

 

If you call "law enforcement" in the city in which these thieves are based, they know all about it.  In fact, they are deluged with angry callers.  I would think that they could do something about it, but I guess that is not high on their list of priorities....

 

There were other glaringly obvious "tells" that these transactions were fake:

 

A). user account almost always has a "Yahoo!" email.  All Yahoo! accounts have been hacked...

 

B). user account on Ebay is "guest" account

 

C). user account on Ebay is very new...less than a week old.

 

D). fraudsters are trying to buy similar things, which are small electronics with high dollar to weight ratio

 

E). Delivery name is different than account holder name.

 

F). Delivery name is almost always FSU in origin (Former Soviet Union).  For example, account holder name is "James White", delivery name is Svetlana Orichinko.

 

G). Ebay user account has zero or 1 feedback.

 

H). and of course, delivery all to the same address!

 

There may be other similarities that I am missing...but it is immediately obvious.  Perhaps the simplest of all, is to not allow any business to be conducted at that problematic address.

 

I had DOZENS & DOZENS of problem transactions in a six week period.  I doubt I was the only one.  I let Ebay & PayPal know about this problem every time it happened.

 

Another example of AI being asleep at the switch is banner ads.  Most of the time, the banner ads that I see are of NO INTEREST at all to me OR wildly inappropriate.

 

Another example is that no computer has passed a Turing test.

 

Or final example is that there are well known intellectuals screaming about how AI is going to exterminate humans.  Well, if you ever find yourself being chased down by a HK, simply go through a doorway and close it.  There was a DARPA challenge for bots to walk up stairs, open an unlocked door, and then walk through it.  Nobody could do it...not even close.

 

So in my experience, I would posit that AI is a long way away.  I think it is being "hyped" and I just don't see it coming, not anytime soon.

 

 

[DooDiligence]

For all Apples' bluster, Siri still can't pronounce some of my friends names right (I've told her again & again...)

 

I lay claim to possession of artificial intelligence.

[randomep]

 

 

But look at Alphazero kicking Stockfish's butt.

[rkbabang]

 

 

But look at Alphazero kicking Stockfish's butt.

 

“How about a nice game of Chess?”

[Jurgis]

Self driving cars are awesome.

 

But if that's "not-real-life" for you yet, then check out the collision avoidance systems that are already integrated into most of 2017-18 models ( https://en.wikipedia.org/wiki/Collision_avoidance_system ).

[oddballstocks]

There are really two ways to look at this:

 

1) True AI - My co-founder and myself joke that getting to artificial stupidity would be a good first step.  We're a long way off here, maybe a step or two off the starting line.

 

2) Situational AI - This is leaps and bounds beyond what was available a few years to decades ago.  This is self-driving cars, or behind the scenes apps.  The intelligence needs to be coded into it, but it's extremely powerful.  This is because there are bounds.  True AI would say that a self driving car might decide one day to learn chess.  But situational AI doesn't need that and doesn't care. 

 

Consider your day to day.  The automatic routing for Google is mind blowing.  It evaluates real-time traffic, looks at estimated future traffic based on patterns and gives you a route.  What's even crazier is there are thousands asking for the same routes at the same time, and Google is intelligent enough to not cause traffic jams.  It is able to re-route people and disperse traffic.  That's incredible.

 

I think AI and robots are two separate discussions.  The intelligence to program something to recognize doors, recognize stairs, and understand how to turn a knob is fairly simple.  The difficulty is programming a robot to mimic human muscles and interactions.  You're conflating two different things here.

 

Ultimately AI will slip into different aspects of our lives.  Like Google maps we won't notice it improving.  I don't believe in the singularity, but I think AI will be helpful and will make people's lives easier.  I know some of the software my company builds might be viewed as AI-ish to end users, but it's really just taking a lot of information, knowing what to do with it and outputting a value.

 

 

[rkbabang]

Artificial Stupidity?

 

“Forever generating unique inspirational quotes for the endless enrichment of pointless human existence”

http://www.iflscience.com/technology/ai-trying-to-design-inspirational-posters-goes-horribly-and-hilariously-wrong/

 

[Sportgamma]

ricky gervais learned totalitarianism from adolf hitler, the inventor of atheism

- Tay, March 26, 2016

 

Whenever I need a laugh, I just look up tweets by Tay the Microsoft AI chatbot or scripts from that AI that was assigned to watch Seinfeld and to write new episodes.

 

Jokes aside, I do recommend this EconTalk podcast with Rodney Brook: http://www.econtalk.org/rodney-brooks-on-artificial-intelligence/

 

It's refreshing to listen to amidst all the AI bullshit being peddled in the (social) media.

[Cigarbutt]

I've been using specialized voice recognition software for three years and have come to the conclusion that there has been displacement of human work and increased productivity.

 

Three levels:

 

1-basic transcription work: level reached.

 

2-first-level analysis: level not reached, at least not quite. A while ago, an experienced and knowledgeable assistant doing the transcriptions would suggest additions and omissions when, for intance, comparing for coherence and completeness as expressed in other previous reports (data base). It looks like this could be "artificially" achieved relatively soon.

 

3-second-level analysis and other higher cognition: IMO sustainable moat will be enduring for a while.

[SHDL]

The WSJ had an article a short while ago on this topic which I thought was very informative:

 

https://blogs.wsj.com/cio/2018/09/07/deep-learning-is-it-approaching-a-wall/?mod=hp_minor_pos6

[SHDL]

Cool stuff on the robotics side (from Japan):

 

[scorpioncapital]

Where is even the natural intelligence :)

 

[Jurgis]

The WSJ had an article a short while ago on this topic which I thought was very informative:

 

https://blogs.wsj.com/cio/2018/09/07/deep-learning-is-it-approaching-a-wall/?mod=hp_minor_pos6

 

It's not bad article, though it simplifies things a lot (for popular consumption). You might be better off with the actual source https://arxiv.org/pdf/1801.00631.pdf that this article summarizes (it goes from 27 pages to half page, so there's that  8)). Although even source article is IMO biased towards issues, but then that's the goal of the author.  8)

[SHDL]

The WSJ had an article a short while ago on this topic which I thought was very informative:

 

https://blogs.wsj.com/cio/2018/09/07/deep-learning-is-it-approaching-a-wall/?mod=hp_minor_pos6

 

It's not bad article, though it simplifies things a lot (for popular consumption). You might be better off with the actual source https://arxiv.org/pdf/1801.00631.pdf that this article summarizes (it goes from 27 pages to half page, so there's that  8)). Although even source article is IMO biased towards issues, but then that's the goal of the author.  8)

 

I totally agree. 

[LC]

Cool stuff on the robotics side (from Japan):

 

 

Current state of robotics (Alibaba warehouse)

 

https://i.imgur.com/MwQiJ5K.gifv

[rkbabang]

AI is trying to find the Zodiac Killer while writing poetry.

 

"Snap my finger on a dinosaurs,

Catch a bullet from a lower jaw!

And when I hear the sound of thunder roars,

Fear of being blinded by the claw."

 

https://www.iflscience.com/technology/ai-programmed-to-solve-zodiac-killer-mystery-creates-creepy-poetry-on-the-side/

 

[rb]

Cool stuff on the robotics side (from Japan):

 

Great! they invented a robot that probably costs like 5 million dollars to do the work that some guy that gets paid 17 bucks an hour can do 5 times faster and better - cause I don't see that thing shimming the drywall.

 

That's real progress!

[SHDL]

Cool stuff on the robotics side (from Japan):

 

Great! they invented a robot that probably costs like 5 million dollars to do the work that some guy that gets paid 17 bucks an hour can do 5 times faster and better - cause I don't see that thing shimming the drywall.

 

That's real progress!

 

I agree that this doesn't look commercially viable (at least not yet).  Same goes with stuff like this:

 

It is quite striking how much better & cheaper humans still are vs robots at performing so many seemingly simple tasks.  I think Chipotle at one point tried to develop an automatic tomato dicer for its kitchens and gave up pretty quickly.  Maybe it is precisely because people can hire cheap labor to do these things that developments in this space have been a bit slow...

 

Anyway, it’ll be interesting to see how things evolve (or not) from here.  I had a former colleague who loved telling people about how his parents used to joke(?) about dishwashers in the 1950s — that they were “pointless” inventions because anyone who could afford one already had servants who would do the dishes “for free.”

[DTEJD1997]

Cool stuff on the robotics side (from Japan):

 

Great! they invented a robot that probably costs like 5 million dollars to do the work that some guy that gets paid 17 bucks an hour can do 5 times faster and better - cause I don't see that thing shimming the drywall.

 

That's real progress!

 

I agree that this doesn't look commercially viable (at least not yet).  Same goes with stuff like this:

 

It is quite striking how much better & cheaper humans still are vs robots at performing so many seemingly simple tasks.  I think Chipotle at one point tried to develop an automatic tomato dicer for its kitchens and gave up pretty quickly.  Maybe it is precisely because people can hire cheap labor to do these things that developments in this space have been a bit slow...

 

Anyway, it’ll be interesting to see how things evolve (or not) from here.  I had a former colleague who loved telling people about how his parents used to joke(?) about dishwashers in the 1950s — that they were “pointless” inventions because anyone who could afford one already had servants who would do the dishes “for free.”

 

Hey all:

 

I have read some science fiction books that are very interesting in regards to robotics...They claim that in the future, robots do the very high level, high value work, and humans still do a lot of other stuff.  Robots are just too expensive.  Why spend $100k on one robot when you can hire a capable, motivated worker for $2/hour in the 3rd world?  Further, if business changes, you can simply let go of the worker, a robot?  You bought it!

 

 

[SHDL]

I have read some science fiction books that are very interesting in regards to robotics...They claim that in the future, robots do the very high level, high value work, and humans still do a lot of other stuff. 

 

I was actually having similar thoughts yesterday as I was writing the above.  That would be a really interesting turn of events.  And like you say, it’s kind of plausible too once you think about the economics...

[rb]

I have read some science fiction books that are very interesting in regards to robotics...They claim that in the future, robots do the very high level, high value work, and humans still do a lot of other stuff. 

 

I was actually having similar thoughts yesterday as I was writing the above.  That would be a really interesting turn of events.  And like you say, it’s kind of plausible too once you think about the economics...

I highly doubt any of that. The thing is that humans are very special beings. They can do low level stuff really well and they can do very high level stuff really well. Humans have things like inclings, hunches, instincts, and sparks of genius and can be invaluable. You're not gonna replace that with a robot. You can't program genius.

 

I remember a line from a show I watched where a guy (Mandy Patinkin) says that when a computer decides to commit suicide because it got too fat that's when I'll believe that computers are gonna replace us - or something to that end. That's a brilliant line and wholeheartedly agree. The source of our greatest flaws is also the source of our greatest achievements.

 

Another thing I read somewhere that I agree with is this: "We'll know that we've created Artificial Intelligence when we make a self driving car and that car decides to take up painting instead". Intelligence means that it must have the ability to think. If something thinks, its thoughts most likely won't go the way that you want. Anyone that ever had a girlfriend knows that.

[Jurgis]

Humans have things like inclings, hunches, instincts, and sparks of genius and can be invaluable. You're not gonna replace that with a robot. You can't program genius.

 

This is a Hollywood trope that has very little basis in reality. I'm sure Chess and Go grandmasters thought they had inclings, hunches, instincts, and sparks of genius too.

 

Yes, there is currently no AGI. This doesn't mean that A(G)I cannot eventually outperform humans in pretty much any area.

[rb]

Chess and go have strict, defined rules that you must follow. That's easy to do.

 

I'd like to see that computer that comes up with the theory or relativity that because it's thinking about riding a beam of light from the back of a bus.

[Jurgis]

Chess and go have strict, defined rules that you must follow. That's easy to do.

 

LOL.

 

It's a standard dismissive response: It's not AI, it was easy, it's only AI when it's AGI, it's only AI if it exceeds human in every single area, etc.

 

Every single time AI does something (better than humans), it's the same response.

Jeopardy?

Poker?

League of Legends?

Image recognition?

Self driving?

Medical diagnosis?

 

I'd like to see that computer that comes up with the theory or relativity that because it's thinking about riding a beam of light from the back of a bus.

 

Not to lessen the value of theory of relativity, but what matters is the theory and not the feel-good story about riding the beam of light from the back of a bus.

 

I'm sure computers will be able to do the theories better than humans (minus the story) pretty soon.

[rkbabang]

"the MCAS appeared to have mistakenly sensed a looming stall and tried to force the plane's nose down. The pilots responded by pulling the plane's nose up to compensate, only to have the MCAS system force the nose back down again. In effect, they were wrestling with software and hardware inadvertently trying to kill them."

 

Boeing 737 pilots battled confused safety system that plunged aircraft to their deaths – black box

Data suggests one of plane's many brains was stuck in anti-stall mode