ChatGPT Is Now a Search Engine. What Kind Of Implications We Might See?

I dont think so…

3 Likes

Exactly!! One of the best AI related quote I have ever seen!

2 Likes

@daniel.m.tripp
Re AI , laundry and dishes

It is coming, we can already get a robotic window cleaner

5 Likes

Imagine the spycam in this, what it could see …

I do our windows, about once a week as we get a lot of sea spray. But i notice that windows have 3 sides, i go outside, clean them, return inside clean them, but the 3 rd side is always a trace i cannot get to.

4 Likes

everyone, everywhere, with AI IoT smart homes constantly getting hacked… good luck with that AI…

Never mind the “Turing test” - never mind how easy it is for AI to “social engineer” - what about a human hacker breaking into AI systems with a bit of “feely touchy” social engineering - mark my words - if it hasn’t happened - it will…

4 Likes

Yes, I get it! :grin:
Just do the very last wipe on one side strictly vertical only, while strictly horizontal only on the other side. The direction of the trace will tell which side it resides :slight_smile:
Well, if it is diagonal, I suspect there’s a fractional dimensional space-time curvature, and cannot reach the 3rd side really, for that a woman’s hand is needed. :slight_smile:

4 Likes

Normally she is the one that notices, but chooses not to labour the point in case I stop doing the job.

1 Like
2 Likes

I don’t want AI to be a teacher, professionally. Because, AI may teach all the subjects with super perfection. But, a teacher’s responsibility doesn’t end into teaching a subject only. Proper upbringing of a student, putting ethics in their minds are also a teacher’s responsibility which an AI cannot do. It is impossible for an AI to FEEL, at least for now and the day when (read ‘if’) AI would start to feel, that day humans would no longer be the master of their fate. The movie ‘Terminator’ will start to become the reality.

To be safe from this type of scam, don’t keep the internet on all the time in your phones, laptops and remove all the access permissions from microphone and camera and on laptops, cover the camera and mic firmly with a tape or paper or just turn them off when you are not using it. But yes, AI technology is scary, specially when there is no legal control on its use.

Yes. AI wouldn’t. But what if, a person, who has no talent for writing (because I think writing is an inborn talent and it cannot be developed by just practicing. It takes lot of thinking, imagining. AI can calculate but cannot THINK or IMAGINE… at least till date) writes a novel using AI and wins a Booker prize? “Drawing is an art, and art is done by humans, not by machines” - this was said by the artists, once. Now see what people are doing with AI. No. It’s not ‘incredible’, it’s disgusting. AI can copy the style of Auguste Rodin, but cannot create the ‘The Thinker’ by its own. It can copy Shakespeare’s style, but cannot create ‘Othello’ by its own. But someday, in future, AI could evolve to a self-aware entity and that day would be the end of the humanity… The Judgement Day .

AI can calculate… cannot THINK on its own… at least till now. I have already written my opinion on this, so I am not duplicating it.

Yes. It will. Because AI is a technology… a programme… a software… made by zeros and ones. So it can be hacked and that hacker would be a human.

1 Like

Exact ! Try to play a chess game against ChatGPT… It will play an impossible move after 5 moves. Then ask where pieces are, it could show 3 kings on board and 2 pieces on same square. Continue questions, the more you ask, the funniest it is.

A 4 years child does better here :slight_smile:

I have a few transcripts, asking a programming challenge, playing chess, …

Everything is wrong.

Yesterday on /. it appears AI solves 2% of math problems… AI Systems Solve Just 2% of Advanced Maths Problems in New Benchmark Test - Slashdot

More ?

2 Likes

Math problems and chess are analytic.
It should be able to solve them… provided its training gives it enough background .

I suspect the training in those areas is deficient

xxxxxxx
NOTE Subsequent discussion has proved me wrong here.
Thank you @bodiccea .
AI is not analytic. It can not deduce anything. It is good at remembering real people’s deductions and can feed them back to you.
xxxxxxx

Programming is not all analytic… there is a creative element … AI is not creative… it cant write a poem or make a movie or come up with a new theory or find a new algorithm.

xxxxxxx
Creative and analytic are different, but AI cant do either.
xxxxxxx

2 Likes

I don’t really agree with your statements… ChatGPT is only LLM…

LLM will never be able to solve a chess position, a math question, or a programming question. I classify the 3 as problems needing deep understanding of a new situation, and needing both knowledge, reasoning, and creativity to solve. LLM does not have any “understanding” of a situation, no reasoning, and, as you say, no creativity.

This is why the term “AI” was not a good term for those models. The “I” in “AI” is simply fallacious. Maybe “AP” would have been better. “P” for “parrot” :slight_smile:

A chess game where ChatGPT castles twice

I thought it could do more than recognise and generate text.
There has to be some interpretation (ie finding the meaning) of the text in abstract terms. Otherwise it cant deduce anything?
Am I wrong?
Is it really nothing more than a sophisticated man page lookup?

Something else I dont understand… I have built algorithms requiring training… you need a training dataset with ‘known answers’
but
AI uses training datasets with no ‘known answers’… just buckets of text with no meaning attached.
How can it learn anything from that?

2 Likes

You are using two terms “deduce” and “learn from”. AI or a parrot cannot do these… You could read “Destination: Void” from Frank Herbert, in which human create a real AI (with a conscience), but are so afraid of it that they “awake” the machine on a vessel which cannot come back… We are very far from that…
AFAIK, ChatGPT cannot play chess (I mean simply follow chess rules, even not being very good), a 4 years old child can do that…
The child did learn the rules (10-15 min). The AI did not, and will “repeat” moves that are impossible (my 2 castlings transcript above).

2 Likes

OK, you have convinced me . AI is not analytic, and it is certainly not creative.
So, if it us able to do say

365  + 132 = ?

it is because it has copied it from somewhere, not because it knows the rules of arithmetic.

I will go back and flag my reply as incorrect… Thanks

1 Like

Yes ! We did learn how to make this addition when we were young…
AI did not. Einstein was able to create his two theories with his brain, no AI could have done that (we could say the same for previous “geniuses”: Archimedes, Newton, etc…).

2 Likes

We learn some things by rote… eg multiplication tables…
but we learn addition by method.
AI might see a description of a method, but it would not know how to apply it.
I am getting it now, thanks.

1 Like

Yes, we did learn multiplication tables… But we can understand they are additions: If you forget what is “8x7”, you are able to use your memory for “8x8, 8x6 or 8x5”, and make an addition or subtraction, even using your fingers if needed…
Archimedes was able to calculate Pi to some decimals, I think you and me could do the same if necessary…
This is where we are very different from any LLM AI. AI will give you Pi with 1 million decimals, but will not even be able to find 3.14 by itself.

2 Likes