Shite.
*gigantic face palm*
I've been interested in this technology for many years, but especially so when I first heard about GPT3 (which ChatGPT is based on) around 18 months ago. ChatGPT is in fact a cut down version of GPT3, and in some ways inferior (some ways better) but obviously has got all the hype.
Anyway, GPT4 is out now (GPT3 being around 2 years old). Things are moving SO fast it's both utterly amazing and actually quite scary all at the same time. We are working towards what is referred to as AGI - Artificial General Intelligence - a term coined to mean when the AI is absolutely as capable human intelligence in all respects. People had theorized this was decades away, but now many experts are saying we'll be there in less than 2 or 3 years. In many respects, GPT3 is already more intelligent than humans, let alone what GPT4 can do.
The "singularity" is the next mind boggling milestone, by which point AI's are so intelligent, they are able to improve themselves in ways we cannot comprehend, and since they are not limited by human biological constraints, from that point on they will accelerate away from us, becoming ever more intelligent geniuses and leaving us like ants in the dust.
That point may be only a few years away as well. Truly mind boggling.
Here's some links of staggering stuff, for anyone interested:
1. The interview the Google engineer had with Google AI engine, Lamda, which led him to believe it is sentient and which resulted in him getting fired:
2. Dr. Alan Thompson - an expert on this stuff - talks about the journey to AGI
3. A link to one of the AI image generation tools, Stable Diffusion where you can enter text, such as "A painting of a boy chasing a red balloon with a backdrop of Paris, painted in the style of the Impressionists" and it creates in seconds, (from it's own "thoughts") this:
View attachment 72376
Stable Diffusion 2-1 - a Hugging Face Space by stabilityai
This application lets you create images by entering a text prompt. You can also provide a negative prompt to exclude certain elements and adjust the guidance scale for image quality. The app return...huggingface.co
4. An interview with another AI Large Language Model, LLaMa=65B:
I didn't say it was. That said, we don't really understand what consciousness is, so taking that lack of understanding and applying it to very large neural networks with billions of nodes and unimaginable complexity, and dogmatically deciding that such constructions are incapable of thinking, is IMO a rather narrow minded view. Not just your view, by the way - I am not meaning to be rude. I simply mean that we may need to start to look at this from different perspectives. We only have experience of biological consciousness so far, and maybe electronic consciousness can arise from different constructs.
I was referring to the supposed singularity you mentioned. I work with this stuff so it's something I can comment on. Neural nets in their current form can't become aware or develop consciousness, it's not possible. Yes, it's possible they might become so complex they might appear conscious in a few years time, but they will still be obeying the rules set in the connections and won't have any desire or original thought. The most interesting development in the next few years is going to something called out-of-distribution learning, where algorithms trained on one task can use that experience in other non-similar tasks, but again, nowhere near achieving a singularity.
Of course I respect your opinion and I am sure you know much more about it than I do. However, my comment was really based on the comments of Dr. Alan Thomson, who I imagine knows as much about the subject as you do, if not more. Maybe you'd like to take a look at the 2nd video I linked above, "Countdown to AGI"?
I watched the first few minutes and honestly, he's chatting shite. There's more chance of Liverpool winning the champions league this season than us achieving his so called singularity any time soon. If you are really interested in this stuff, the guy to watch is Yoshua Bengio. He's one of the godfathers of modern AI and is a particularly brilliant communicator.
You should watch his video instead of the uninformed slagging him off. He is not some know-nothing Youtuber and cites plenty of references.
Chippy, I watched enough of his video to know he was talking shite and just citing references doesn't mean you are correct.
Can't wait to take mine down the local
Whatever mate. My personal view is that you are behind the times with your thinking, but if it's a hill you want to die on, I am not going to stop you.
If she'd just turn off the two clever buggers in here I'd be happy.
Which, oddly enough, is true for many an AI generated paper…