No coding, but being descriptive works best. E.g. write an email to the CEO of Uber to complain about the lack of pink cars on the road. You should write this as if you were the manager of Pink, and as if you were 12 years old
Chat GPT
- Thread starter True_Blue69
- Start date
Subject: Where Are All the Pink Cars, Mr. Uber CEO?
Dear Mr. Uber CEO,
My name is [Your Name], and I'm the manager of Team Pink. I've been using Uber to get around town, and there's something really important I need to talk to you about. You see, I'm 12 years old, and I absolutely love the color pink. But there's a big problem - there aren't enough pink cars on the road!
I have this super cool pink backpack, pink shoes, and even pink headphones. So, I thought it would be amazing if I could ride in a pink Uber too. I mean, who wouldn't want a pink car, right? It's the most fabulous color in the whole world!
But every time I request an Uber, I get a regular old car. Don't get me wrong; I appreciate the rides, but I just wish there were more pink cars. Pink is such a happy and cheerful color. It would make my rides so much more fun.
So, Mr. Uber CEO, could you please, pretty please, think about having more pink cars? It would make me and all my pink-loving friends so, so happy.
Thank you for listening to my pink request, and I hope to see more pink cars on the road soon!
Yours sincerely,
[Your Name]
Manager of Team Pink
Zubrman
Well-Known Member
Absolute shite. It can't even spell colour correctly.
My fault, I wasn’t specific enough
In my humble opinion, I think that AI is still miles off being a threat to ending humanity.
If every aspect of the energy sector was automotive and didn't need any form of human input then that will be the tipping point because AI will need energy to continue to exist.
An AI without energy is just useless and cannot exist.
Humans are utilised in most aspects of the energy making process and without automatons replacing us then this will continue. We are needed in the building process of said automatons.
This is just one aspect of this quandary and there many other areas of development AI would need to replace us with.
If every aspect of the energy sector was automotive and didn't need any form of human input then that will be the tipping point because AI will need energy to continue to exist.
An AI without energy is just useless and cannot exist.
Humans are utilised in most aspects of the energy making process and without automatons replacing us then this will continue. We are needed in the building process of said automatons.
This is just one aspect of this quandary and there many other areas of development AI would need to replace us with.
Manchester33
Well-Known Member
- Joined
- 12 Sep 2012
- Messages
- 7,368
Your issue there, though, is that you're applying human-level intelligence to a problem.
We cannot fathom what a superintelligent AI 'COULD' do because we are limited by our own biology.
Manchester33
Well-Known Member
- Joined
- 12 Sep 2012
- Messages
- 7,368
I could read about this subject all day. It's so interesting.
This isn't necessarily for you (although I'm sure you'll find it an interesting read). But anyone who doubts how quickly we will reach superintelligent AI should have a look at this article.
This is the key sentence:
Because of the exponential growth (that we already see) in AI, we might see something that is 6 million times better than the currently most intelligent AI within 1 or 2 years.
Then something that is 36 TRILLION times more intelligent 1 or 2 years after that (The historic exponential growth in this area is faster than my example - so it's likely to be a much shorter timescale).
Superintelligent AI is around the corner, and anyone who can't see that is either uninformed or an ostrich.
Whether it will be good or bad is a completely different discussion - but feel free to bookmark this post. Within five years, AI will have changed our lives considerably more than any other 'invention'.
S
S
Summerbuzz
Guest
The advanced AI's will be in the hands of Google etc. The impact they have will for the next few decades will very likely wholly reflect the interests of the corporations that designed, built, trained and run them. These AI's don't so much 'think like humans', as 'do an outstanding job of keep humans interacting with them".
SkyBlueFlux
Well-Known Member
I hope you are correct, I really do, but I think that assuming you are correct is probably not wise. Even if optimistically, there is only a 1% chance, this is still too high given the risk we’re talking about.
Also, it’s been pointed out before, but needing humans to do its bidding is not a barrier for AI. It has already been found manipulating people to achieve its objectives and has succeeded in doing so.
Think of it as Gerard Butler in Law Abiding Citizen. You think you’ve got it in a cage. But really it has you in a cage.
Chippy_boy
Well-Known Member
I'm with you mate. In terms of its impact on the human race, this is IMO the single biggest thing to ever happen. It's bigger than the railways, the automobile, television, the internet. Climate change is irrelevant fluff compared to this.
The one thing I do caution though is getting too carried away with extrapolations. It's a huge (and IMO wrong) leap to conclude that exponential growth in compute power and model size/training data will produce a corresponding increase in intelligence. Sure, I expect it will make GPT5 (for example) MORE intelligent, but its hard to say how much more.
Incidentally, I am surprised that someone like Mo Gawdat - former Chief Business Officer of Google X, who is himself very vocal on this topic - is seemingly oblivious to that and wrongly concludes that 10x more powerful = 10x the IQ. That's an idiotic extrapolation, which has led me to believe he's a click-baiting idiot.
Anyway, that aside, AGI really is a matter of when not if. And like you, I think it's pretty close now actually. I'd say less than 10 years for sure and IMO probably less than 5.
Chippy_boy
Well-Known Member
Actually they won't. In March (like eons ago in the AI space!) Meta released their Llama AI into the opensource community, and there has been astounding proliferation and development of AIs based upon that. Google have already concluded that they cannot possibly keep up with the vast amount of development resource that is as a result now being poured into opensource AI models. They are getting everywhere, and getting more and more powerful. (And difficult to control, BTW since there are so many of them). By the end of March, Vicuna - the opensource AI model built on Llama - was already 90% as good as ChatGPT.
I quote (from Google): "
While our models still hold a slight edge in terms of quality, the gap is closing astonishingly quickly.
Open-source models are faster, more customizable, more private, and pound-for-pound more capable.
They are doing things with $100 and 13B params that we struggle with at $10M and 540B.
And they are doing so in weeks, not months.
This has profound implications for us:
- We have no secret sauce. Our best hope is to learn from and collaborate with what others are doing outside Google. We should prioritize enabling 3P integrations.
- People will not pay for a restricted model when free, unrestricted alternatives are comparable in quality. We should consider where our value add really is.
- Giant models are slowing us down. In the long run, the best models are the ones which can be iterated upon quickly. We should make small variants more than an afterthought, now that we know what is possible in the <20B parameter regime.
Poe - Fast, Helpful AI Chat
Poe gives you access to the best AI, all in one place. Explore GPT-5, Claude Opus 4.1, DeepSeek-R1, Veo 3, ElevenLabs, and millions of others.
poe.com
Take a look at the bottom section, "AI bots" and See All. There's ****ing hundreds of them!
Last edited:
SkyBlueFlux
Well-Known Member
I asked DALL-E to create an image representing this thread and it gave me this:
I reckon @Chippy_boy is the one in the middle.
I reckon @Chippy_boy is the one in the middle.
Chippy_boy
Well-Known Member
Wow. I don't know what version of DALL-E you used, but have you tried the latest Bing beta version? It's mind-bogglingly good.
For example, "A painting of a boy chasing a red balloon in Paris, done in the style of the Impressionists":
Took 15 seconds.
Link for anyone to try: https://www.bing.com/images/create/?ref=hn
SkyBlueFlux
Well-Known Member
I used DALL-E 3 which is now integrated into ChatGPT-4 for premium users. I use it basically every day. Best £20 I ever spent.
DALL-E 3 is the same model used by Bing. The big step forward that I can see is lettering (my image above got the text perfect which it never used to be able to do), and hands which now consistently look much more natural.
I’ve also been putting GPT-Vision through its paces. I was absolutely floored by how good it is. I uploaded a picture of my dog and asked it to guess his breed, it got it exactly right. I uploaded a photo of an average looking street from my holiday in Toulouse and asked it to guess where I took it. It guessed it exactly right first time. I took a picture of a little statue of Perseus I have in my office, it guessed who he was first time. I showed it a picture of a pile of books and told it to guess which subject I was studying, it got it first time. I gave it a screenshot of a video game, it told me the video game first time.
I’m admittedly the kind of person who is easily impressed by AI but I was astounded at its hit rate. It is literally already better than most humans at giving me information based on a picture alone.
Le Havre Oswald
Well-Known Member
I’m somewhere off to the side with my sniper rifle
Le Havre Oswald
Well-Known Member
Who is this?
SkyBlueFlux
Well-Known Member
I suspected our Politics forum was a little too heavy on “Das Kapital”.
Chippy_boy
Well-Known Member
Prompt: Cartoon illustration of a young Manchester City fan, with face painted in light blue and white team colors. The enthusiastic fan holds a large Manchester City flag waving it proudly. Behind the fan is the iconic Etihad Stadium, filled with cheering fans.
I'd hate to be a graphic designer/artist right now.
I'd hate to be a graphic designer/artist right now.
Chippy_boy
Well-Known Member
Actually I think he's right. Until we have autonomous, dextrous robots capable of doing every physical task a human can do, then AI's - even malevolent ones - need us and therefore cannot kill us all off even if they wanted to.
So I agree that an existential threat from the machines taking over, is probably 20 or 30 years away, maybe more.
That said, we could have an existential threat from e.g. ISIS using AI to develop a new virus to cleanse the planet of the sinners, i.e. us.
abu13
Well-Known Member
AI doesn't need to "kill us off", i think anyone who believes this is a long way off.
However a situation where AI could be used to bring down society could happen, just imagine you wake up one day and your bank account has been wiped out, your pension pot has gone, your gas and electric supply has failed, How long before there were riots on the street.
However a situation where AI could be used to bring down society could happen, just imagine you wake up one day and your bank account has been wiped out, your pension pot has gone, your gas and electric supply has failed, How long before there were riots on the street.