Digging in to AI

[JoeWras]

. In the late 90s, my aerospace megacorp looked into applications of speech recognition for use in an aircraft cockpit as another means of pilot's inputs to the flight system. It worked so poorly that people laughed and said it would surely misinterpret voice commands and get everyone killed. I wonder if they have tried again.

When it was cancelled, did you all start chanting, "Pull UP! Pull UP!"

 

[imoldernu]

That's sci-fi. To me, the likely scenario is some key AI that has taken over important human tasks will malfunction due to AI.

Boeing 737

 

[kgtest]

As my man Woz says (Steve Wozniak) the true AI test is when a robot can walk into any house and make a cup of coffee (provided the ingredients and objects required exist).


I think that would be the "holy ****" moment for most humans. Wait a sec, a human never needs to make another cup of coffee...bye bye barista.

 

[kgtest]

Explainability is huge. For the lawmmakers and lobbyists. Why did the machine learn and then produce that outcome? What were the variables, decisions and rules used to determine that Joe Blow gets Bail but Ho Hum does not, or why She Sally is the better recipient of that heart transplant than Him Harry, for instance.

 

[Sojourner]

As my man Woz says (Steve Wozniak) the true AI test is when a robot can walk into any house and make a cup of coffee (provided the ingredients and objects required exist).

I think that would be the "holy ****" moment for most humans. Wait a sec, a human never needs to make another cup of coffee...bye bye barista.

By that standard, "true AI" basically already exists. There are robots that can competently walk and navigate around things, and there are also machines that can make a very decent cup of coffee in an automated fashion. There's not much doubt that a "walking coffee-making robot" could be fashioned today if the right technologies were stitched together into one device.

To me, though, that's not "true" or general human-level AI. To know if you have that, you'd need a much more challenging test. Something like the Turing test: If you, as an external observer, can't tell which participant in a long running, broad-ranging conversation is human or an AI system, then the AI system possesses general, human-level intelligence. There are AI systems currently that can engage in limited dialogue with humans, but none that are remotely close to passing the Turing test in a general way. That certainly won't be the case forever, though.

 

[cbo111]

I just watched a TV interview with a medical doctor this morning praising the coming virtues of AI and doctor visits. While it seems counterintuitive, the rapid refinements of AI will free up your doctor's time-crunch to better interact with you, which will improve the person-to-person contact. They will actually look at you during the appointment, and not their computer screen or file folder. And they can spend more time per patient because there will be no need to transcribe their notes into the file afterward. The "machines" will take care of that, as well as offering useful diagnostic guidance to the doctor.

 

[imoldernu]

So, now we're digging down...
Gotta get from "automatic intelligence" to "artificial"... the REAL HAL.

Who will need a doctor, and the doctor's education, when, not only will health be determined by machines? Not enough? How about no need for machines, as health will be perfect because of atomic/gene manipulation. No sickness, no disease. Ooops...!!!! did you say broken bones? Certainly not with advanced danger recognition.

Yeah... like how do we really know what happened on Mars 10 million years ago?

Sheesh! Remember Albert E.? He was still alive when I was in college. How far have we come in that short time? Now, as we deal with quantum leaps of science, we may see the ultimate... AI... before my great grandchildren even get into grammar school.

 

[NW-Bound]

The early applications of AI were so called "expert systems". The medical diagnosis AI is one of the obvious applications. It is quite useful, and will not miss some pertinent factors, or make dumb errors like a harried doctor would.

I would think that such an expert system in a defined field would be easier to build than a computer program that can play Jeopardy, which IBM already did.

But a medical expert system by design does not spend its spare CPU time to ponder about the meaning of life, the effect of Quantitative Easing and its ramifications, or debate the merit of an air fryer vs. the older convection oven, let alone worry about the developing hurricane that may cut the electrical power to itself, causing it to shut down.

It will be a while yet before a machine becomes a sentient being, and achieves self-awareness.

 

[Sojourner]

Now, as we deal with quantum leaps of science, we may see the ultimate... AI... before my great grandchildren even get into grammar school.

From everything I've read and heard on the subject, superintelligent AI is unlikely to be realized before the middle of this century. I'd guess your great grandchildren will have left grammar school far behind by then (if it happens around 2050). Now, as for your great, great grandkids... different story. Superintelligent AI will almost certainly be a part of their lives. Many (if not most) AI researchers predict a 90% or better chance we'll have the "ultimate" AI by 2100.

 

[Sojourner]

It will be a while yet before a machine becomes a sentient being, and achieves self-awareness.

I agree, but self-awareness is not thought to be a necessary condition for general AI. Certainly, it isn't necessary for specialized AI, such as software that plays chess or drives cars or reads an X-ray. From what I understand, self awareness (or consciousness, or the ability to observe one's mental states) is thought not to exist, or to be very minimal, in animals such as fish, frogs, and lizards, yet they all have quite sophisticated brains capable of solving a range of problems. They have some level of "general" intelligence, IOW, but little or no self-awareness. For some reason, consciousness evolved in animals like birds and mammals, but it's an open question as to why this happened and whether it was simply a natural, emergent feature of the incredibly complex neural anatomy of these animals. Truly fascinating subject, IMHO.

 

[NW-Bound]

No, an AI computer or program does not have to come anywhere near the level of a human mind to be useful.

But it appears that the general public often shows the fear that current technology is on the verge of producing machines like HAL in the movie 2001: A Space Odyssey.

Here's what HAL, the homicidal computer, says when it (he?) observes that it (he?) is being deactivated by having its memory or CPU modules disconnected.

HAL: "Look Dave, I can see you're really upset about this. I honestly think you ought to sit down calmly, take a stress pill and think things over.

I know I've made some very poor decisions recently, but I can give you my complete assurance that my work will be back to normal. I've still got the greatest enthusiasm and confidence in the mission. And I want to help you.

Dave, stop. Stop, will you? Stop, Dave. Will you stop, Dave? Stop, Dave. I'm afraid. I'm afraid, Dave. Dave, my mind is going. I can feel it. I can feel it ..."

The movie was made in 1968. This HAL computer became operational on Jan 12, 1992, and the space mission was in 2001.

We are now in the year 2019, and AI is nowhere at this advanced level.

 

[Koolau]

I have no idea where AI is headed, but my best guess is: Most of the predictions we make about it today will not be (completely) realized in 20 years. Almost certainly, we will be very surprised when we compare today's predictions to reality in 20 years.

If you've ever watched one of those "predictions from 1950" or similar, they get most everything wrong (flying cars come to mind.) Why would we expect AI to be any different?

I think due diligence is in order when we work toward a world (possibly) dominated by AI (see my tag line - again.) Bit with the hope that we will STILL be smarter and more perceptive (maybe not faster) than AI creations, we will probably not end up with SkyNet or similar - let us hope. As always, YMMV.

 

[robnplunder]

But it appears that the general public often shows the fear that current technology is on the verge of producing machines like HAL in the movie 2001: A Space Odyssey.

We have Hollywood to thank for that. How many movies are about AI vs humans (who almost always triumphs - yeah!). But if AI ever advances so much that they are self aware and think and do like humans, I am afraid it's game over for human race. No Hollywood ending there.

 

[target2019]

SwiftKey just let me know she saved me 10000 strokes. Thanks!
Google Assistant noticed I unlock the phone a lot at home. "Do you want to leave phone unlocked while at home?"
I am only as smart as AI allows. Not very, I know!

 

[Bruce]

Google's DeepMind dIVISION is doing some amazing things with AI - they have self-learning Neural software application - read their white papers to see what their flavor of AI

 

[imoldernu]

Try take a look on Google for CES 2019 AI or CES 2020 AI .

More than you wanted to know, but my bet is within 10 years... and after that I won't care.

 

[sengsational]

Talking about artificials taking over, anybody watching "The Orville"? I don't know what to think about that show but keep watching for some inexplicable reason. Anyway, yet another biologicals against machines two part episode aired recently.


I think everyone our age should be hoping for the singularity sooner rather than later. If it turns out to be gray goo, we lived most of our life comfortably. But we could get to place our consciousness in a container and keep going, imagine the marvels of new stuff to discover!