Does AI make you concerned?

How many people actually owned horses? :racehorse:

There may be fewer horses but price for a horse surely went a lot higher now.

Because again it’s all based on llm, it’s little more than auto complete. It doesn’t actually know anything, it just compiles an answer based on how others will statistically respond. It’s little more than an automaton.

We don’t even quite know what consciousness is, let alone try to emulate it.

A giant training model for llm isn’t going to result in an agi no matter what.

If it starts saying, without any prompting, if it has a soul, then we might be getting somewhere.

1 Like

Sucks to be derivative.

The horse population dropped by 80%. All those new jobs created by cars were in glue factories

2 Likes

I expect a lot from AI that is trained to actually understand how things interact in the real world. AI trained to be good at math goes a little in that direction.

The human brain has many specialized regions and parts. In the future we may have an AI composed of many individual modular models that through specialization, synergy and increased complexities then become closer to being really intelligent. Compute is the bottle-neck I think.

1 Like

Has anyone used Copilot for M365? A colleague posted a summary of last 5-10 minutes of a Teams meeting yesterday and it was extremely uncanny to have the AI summarize what we had been discussing and even have it accurately describe how I had agreed or what I (and others) had suggested on various topics. Sure, AI knows a lot about a lot of public data, but to have it accurately make statements or generate new insights about what was discussed in a closed meeting really impressed me.

3 Likes

Yes, I think genuine human talent may become more valuable.

We generally don’t buy things in a strictly rational way. A lot of how we feel comes into it, hence marketing.

“That Human Touch” ™ might, eventually, hopefully, become valuable

Yea, its annoying that AI became the buzzword when most of it is more to do with to algorithms.

Computers are already good at math, they’re designed for it. Math is inherently logical, and so don’t need ai to solve.

Computer used to be a job title for math geniuses who solved math equations and made tables for trig and other irrational numbers.

Then we built machines that did it automatically and used it to make tables for artillery pieces so they can hit their targets (the first computers did mostly that).

And how is this a problem? What is the horse equivalent to AI taking over like cars did?

AI is already “solving math” in ways human literally cannot think of.

1 Like

Well… yes and no. Algorithms is involved in deep learning in what and how to give feedback to an AI model to help it learn. Then there’s model architecture and you try to think of an architecture that might do better, but nowadays most of it is trial and error or letting AI to design the next-gen architecture.

The actual learning is all done by the artificial neural networks. Teaching a deep learning model to read and write is not unlike how we learned how to read and write. You just make it read a lot of articles and then ask it to write something, then you grade how well it did and it figures out what it did wrong and how to improve.

Well, what will be interesting to watch will be two independent AI systems self learning and then learning from one another in a language we cannot understand.

But, still be at our beck and call. :phone:

Now imagine if you try to shut off an ai or delete it, it screams for help and pleads with you to stop…

Or it starts a nuclear war because it doesn’t want to die.

I believe it’s possible now. At first all the weights in these artificial neural nets are randomized. If we use two of them to feed one another inputs and then give the loss to the other’s output (grade the results), they will eventually converge at something they can mutually understand, which would be utter non-sense to us.

I’d like it to simply delete ignorant claims
and worries from the internet. 🛜

Some people are ruining it for the rest of us.

1 Like

Without rule-based filters, I believe AIs can already do that.

Yes, but what task has it been given? Solve math or figure out how one family can get out of generational poverty?

I entered the following prompt, and the AI wrote the lyrics, the music, and sang it.

Ko criticizes Lai’s family home of being an illegal building, when it is actually the result of an injustice suffered by the miners living in the area. Ko exposed to own illegal buildings himself, including a paved over farmland.

2 Likes