Yay... A new Dylan Beattie talk! And it's great as usual!

Having books written by LLM and not have them to be declared that they are written by LLM seems like a huge problem. Interesting and enjoyable talk!

@kalmarnagyandras

4 ай бұрын

The bigger problem IMHO is the amount of good reviews it has... :)

@user-bk9wt8gk1q

3 ай бұрын

@@kalmarnagyandras Probably also bots? 😅

An amazing talk making great amount of extremely valid points

Count on Dylan to have a measured take on the inflection point of generative AI. Great stuff.

44:00 I do really like this idea. That machine intelligent systems will become “legal person” long before consensus of if they are a person or if they are sentient is reached. Like, if it can walk like a duck, quack like a duck, and can run a company better than a human could… then yeah. Imagine, “SASHA is the Chief Operational System of Sasha Hedge-funds, and she owns 60% of the corporations stocks”.

Strong Rabbit R1 vibe with the proposition at 50ish minutes. Great talk, as always, thanks !

I still remember when VR was poised to totally transform the way we do everything.... and, for that matter, at least 2 times when "proper, real AI" was just around the corner and was about to put all the doctors and lawyers outof business. (Well, one time at least)

51:15 social network where you can add friends only by touching your phones

Jeez, I haven't seen a drum scanner in nigh on 20 years… ironically, while I was still back in Brisbane.

Excellent video. You make a great point at the end about using AI together with humans. But as for DALL-E 3 getting the text wrong, patient persistence has worked well for me, by either telling it several times in the same conversation what it spelled wrong or starting a new conversation completely. This only makes sense if it would take you a long time or impossible to create the image manually. You obviously have more talent than I do with graphic design.

In order to gain more acceptance of Artificial Intelligence in companies, the management should be replaced by AI first.

I just rewatched this, and as you're going on about the British Airways chatbot.. Air Canada tried that and the chatbot's (false) advice was in fact found legally binding for Air Canada kzread.info/dash/bejne/kZiJ1pluhLO3gMo.html

Chat-GPT also gaslit me when I asked: "What is the smallest even number greater than 2 that cannot be expressed as the sum of two primes?" It came back with 4. Then apologized, and said 6. Then 8. etc... And yet when I asked it what the Goldbach Conjecture was, it answered correctly and told me it had been verified up to a very high number (I forget how high exactly, but much higher than 8!)

The potential for AI replacing humans is less of a "all X are replaced" and more of a "we only need half of our X because they can do twice the amount of work". Developers in general will not disappear but we will increasingly do more and more multiples of the amount of work we do now and have fewer and fewer peers.

@crism8868

Ай бұрын

That's a recipe for burnout in my book 😐

I think there's another tipping point coming, caused not by technological innovation, but by unrestrained greed: Nowadays, it's hard to buy music. Mostly you rent it. And so physical media is becoming popular again because people like to own stuff. (Though, you don't need physical media; you just need to be able to download music and listen to it without an Internet connection... which the big streaming companies hate.)

20:00 give the teledildonics people some more time

A thousand dollars for a webcam?!?

Who's building APIs designed to be queried by LLMs/APIs that can take user garbage prompts and figure out what they actually asked for????

bullish on original thought and authenticity.