We can put it to bed, AI is stupid.

quote:

Originally posted by Patriot:
It's no different than liberal parents raising a liberal child with liberal views.

Whomever programs the AI is the "parent" and can therefore program in the parents bias.

Bingo right there.

quote:

Originally posted by architect:
How many AI's do we have here on SF? Some of you guys I'm not so sure about....

Joking aside, I agree that this is a very significant development. I recently had a CT scan performed where the results came back so fast (minutes) that I am convinced that they had to have been generated by an AI rather than a human doctor. This didn't seem to matter to my gastroenterologist.

When AIs can reliably and unfailingly pass a Turing test, it raises a lot of questions. Can they be considered intelligent beings? What rights do they have? As a defender of human supremacy, how do I determine that I am talking to another human rather than an AI? If AIs are prohibited from participating in certain activities, how would you tell?

If a machine is talking to you, it's not talking to you, machines can't talk to anyone, they're machines. The problem will be when the people assign machines as the final arbiter. It won't be skynet but it will suck real bad. Imagine your criminal or civil case being decided by an AI. That would be worse than an evil human judge.

quote:

Originally posted by maladat:

quote:

Originally posted by Georgeair:
Most of what folks are calling AI now is really just machine learning rather than AGI. Very different.

As far as stuff that is actually happening and getting discussed in popular media, machine learning is pretty much it.

AI-as-a-subfield-of-Computer-Science includes a bunch of other stuff, but the other stuff isn’t dramatic and exciting to the average non-Computer-Scientist and gets very little public attention.

AGI/strong AI is still very firmly in the realm of science fiction.

Agreed. Sabine Hossenfelder has a youtube video describing AI/MLs as becoming potentially good tools for very specific applications, and I agree. But as a benevolent general purpose intelligence, not really. Humans will create ML/AIs to harass their competitors and enemies, and that could become a problem. They already have done it in a way when you look at what the vast left wing conspiracy has done with Twitter for example.

After I watched this Netflix movie, I wanted to uninstall all Social Media. But machine learning is really dangerous and there is little understanding much less control for this tech.

Perfectly normal. No need to be concerned.

Microsoft AI chatbot threatens to expose personal info and ruin a user's reputation

Combine this kind of evil, aggressive attitude with robots that can see in the dark, run faster than you, are stronger than you, think quicker than you, and are programmed by people who want to kill you.

Nah, no problem at all. The future looks bright.

quote:

Originally posted by Skins2881:
We can put it to bed, AI is stupid.

damn,

I heard that in the characters voices from the movie,,,,



meanwhile,

Black Mirror, with the exception of the first 3 episodes,, one of the best series I have watched,


those dogs,,, shot this at Shot Show last month,

https://www.youtube.com/shorts/85ZwMSydXwc

not sure if Ball will stop one, but guessing AP will

quote:

Perfectly normal. No need to be concerned.

I like my AI spicy, HAL was way too boring.

I'm more in tune with AI/machine learning than than the majority of people on here. They have to power to cure cancer, allow an interstellar presence, and/or kill most humans. It will be transformational for society to say the least.

I'm hopeful for Data (Star Trek) vs Skynet, but either way you'll never new to worry about if you have the new cover page on your TPS report.

quote:

Originally posted by parabellum:
Combine this kind of evil, aggressive attitude with robots that can see in the dark, run faster than you, are stronger than you, think quicker than you, and are programmed by people who want to kill you.

Nah, no problem at all. The future looks bright.

A few years ago when Boston Dynamics was demoing their robotic dog Elon Musk warned of 80 MPH, well armed, smart robots. Seems like the software and the mechanical engineering are getting very close to what you and Elon are talking about Para. But hey, me worry?

quote:

Originally posted by wrightd:

Imagine your criminal or civil case being decided by an AI. That would be worse than an evil human judge.

I'd take an unbiased AI that applies the facts and law over judges who have no regard for either. Will it be unbiased... that's the million dollar question.


I can't wait for AI to put a lot of these smug white collar types out of work. When blue collar jobs were getting decimated the popular phrase was "learn how to code/program." When AI eliminates 90%+ of those jobs I can't wait to see what sitting behind a desk staring at a computer screen making 6 figures qualifies them for in the real world.

quote:

Originally posted by Hildur:
I'd take an unbiased AI that applies the facts and law over judges who have no regard for either. Will it be unbiased... that's the million dollar question.

Oh my. Shades of Donald Fagen:

"A just machine to make big decisions
Programmed by fellows with compassion and vision"

Programmed by fellows with compassion and vision. Aye, there's the rub.

Just one question for you- do you prefer lubricant, or do you like it dry? Because if you let machines decide your fate, the only question is how you will get fucked in the ass, and not if.

You can't get away from the human element, even with these machines, and you are placing your fate into the hands of something which is not even alive, and that is insanity. You would be better off by having a slot machine determine your fate. Or perhaps the flip of a coin. That would be the "green" solution. The only carbon emission would be when you crap yourself on the way to prison.

This obvious satire vid always makes me laugh

https://m.youtube.com/watch?v=y3RIHnK0_NE&vl=en

I don't hunt elephants, but I've got a 416 Ribgy and a bunch of solids. Gotta have something that can punch through that titanium combat chassis.

There are people that think they want to work on A.I. projects and a week later after learning the very basic aspects and applications,
They back out and disavow any and all connections.

The lack of redundant potentiometers or on off switches is the major concern.

Nondisclosure agreements are are a huge deal.

My apologies if this link has been posted in another thread.

It seems as though the folks at AI are improving the learning abilities of the monster they have created. Chat GPT-4 the latest version has improved in visual recognition and comprehension. It also learns all sorts of stuff much faster. The article is quite long so I'll post some highlights. It is definitely worth the read an researching what is stated in the article. ChatGPT 4 is available to the public. What is happening in secret research centers gives me a lot of pause.

Side Note:
In my professional experience the team I lead is using ChatGPT for some of the work we are doing. Currently we are porting a very large legacy system from C++ to C Sharp. Even though both languages have a C in their name they have very little in common. We will paste the methods written in C++ and ask the bot to convert it to C Sharp. Very accurate and fast. Chat GPT takes seconds to perform some tasks the best developer on the team would take hours to do. On some conversions "judgement" is required on the approach to take for the conversion. ChatGPT seldom exhibits bad judgement and when it does it will learn if you tell it why you don't like the approach it took.



Back to the article:

* The Uniform Bar Exam - 298 our of 400. Passed in most states

* The SAT - 678 our of 800

* GRE - 99th Percentile Verbal, 54th Percentile Writing

* USA Biology Olympiad Semifinal Exam - Scored in the 99th Percentile

* AP Exams - Recevied a 4 or above on some of the exams (Art History, Biology, Microeconics, Physics, Calculus, etc. )

* AMC Exams - 45th - 66th percentile

* Sommelier examinations (win tasting) Average of about 85% on all three exams

* Wharton MBA exam - Recieved a B or B-

* US medical licensing exam - At or near passing levels

* Essays - Wrote very good essays

* Microbiology quiz - "Nailed it" by Phd giving the quiz

* Law School Exams - "low but passing grade in all four courses" approximately equivalent to a C+

* Stanford Medical School clinical reasoning final - ChatGPT passed a Stanford Medical School final in clinical reasoning (72%)


Link

I had no idea it was developing this rapidly. I was reading the AI Art thread started by Skins. From there I have been down the rabbit hole reading and watching the dangers of AI.

I’m kind of shocked at how rapidly this has developed. The scariest part to me is the exponential growth. Between that and humans being too foolish to believe that a machine will be smarter than them… this isn’t good.

quote:

Originally posted by Beancooker:
I’m kind of shocked at how rapidly this has developed. The scariest part to me is the exponential growth. Between that and humans being too foolish to believe that a machine will be smarter than them… this isn’t good.

The exponential growth is what makes me think we should stand back and evaluate (we won't). In just three months of seeing how much AI improved at porting code from one language to another seemed at first awesome, then very troubling. The thing was learning. I had not paid much attention to the art that could be generated but that too gives me pause. It is capable of rendering astounding art pieces. Add to that visual recognition improvements leads me to believe that this technology could go south with or without human intervention. I also have to believe that the "improvements" in it are aided by AI itself. Think about that, if it is assisting in improving itself, will it be able to do improvements on it's own? Some think not. For me that is a huge question mark.

That’s exactly it. Exponential. A true understanding of how fast it really is. People just don’t understand that, as most things in the world do not have exponential growth. In many instances where there is, people fail to see it as such.

This is also to your point Bytes; intelligence. If a robot has the same five senses we have, but can see in thermal, as well as millions of shades, from -200 to +200 with crystal clear vision, smell that can analyze what is in the air, from pollens to gases, to who knows what. The ability to hear everything within a huge radius and determine the different conversations and sounds.
If a robot is capable of all that, (we currently have the technology for everything I listed) what do we not know about? When the AI has communicated with humans for many years, and can understand our patterns and the way humans interacts it will be able to truly understand us. When that happens, it’s becoming more intelligent than we are.
Look at it converting C++ to C Sharp. As it processed more code for you, it became better at it, making fewer mistakes and changing things to perfection. Things as you stated, would take a senior programmer hours. If it can learn that at an exponential rate, imagine its ability to code after it is used for much more complex coding.

So you now have a robot that is far superior to a human in every physical capacity. It runs on a battery and is quick charge with solar capabilities and probably some technology we will soon learn about. This superior bot has a supercomputer telling it what to do, and in essence, it can perform tasks and operate independently from the supercomputer, while still carrying out the duties it was asked to do, but absorbing and learning and gathering intelligence to report back. iRobot really might not be as far off as people think.

I know it seems like fantasy, but seriously think this through. In 2001 I had a cell phone. It made calls and it was a large POS. The thought to surf the internet, and what the internet has become, the ability to have a video chat like you’re almost really in the same room, the fact that my phone is exponentially more capable than my 2001 phone goes to show how quickly things can happen.

My analogy covers a span of 22 years. My point to that is the old “penny a day doubled every day for a month”. Proving exponential growth.

When/where in that month are we now? Are we at the fifth day, tenth, twenty-ninth? The growth won’t stop at a month… what then???

AI WILL make it to the battle field.
Just wait until AI drones are dropping grenades in coordinated waves against targets identified using IR cameras.
And directing mechanical infantry like to robodogs.