Fair questions in any age, save the technology related ones. Most have asked the same questions over and over. Some answers have lead humanity further than where it was before. So good for you to keep moving the wheel of change.
Spoiler alert (not that much): change is coming. Always.
You guys are so much really sure all life needs to take off from chemical reactions.
I wouldn't be so sure. Not even for our Sun system, nor even our own planet.
Humans really, historically have had little imagination, think electricity (THE big one we missed as a species for literally thousands of years and it was 100% of the time all around us), wifi networks in pre 70s scifi, etc.
So it could be happening again, we are again so much sure we know everything, "this is it", "the peak of science", etc.
my comment before knowing the information you provide in the text is simple, these models are trained over some texts downloaded from Internet, there are some real "trends" you can easily spot reviewing yourself some subreddits, i.e. likely many american citizens commenting in a guns related subreddit have certain opinions which probably aren't shared for other people in a LGBT+ subreddit elsewhere.
It could be matter of choice during the assembly of the datasets used for the training, simply. They could have included some subreddits comments from here and there, but some richer texts were found in some subs, so they included more text from there, hence the LLM ended having "biases" related to those comments, because some ideas are more "heavy" in their internal values than the other ideas from other people, which are less - stadistically -represented in the dataset.
It looks like you could really introduce any "bias" you choose into the "core internals" of any LLM. This could be an accident or not.
No, not really. Yes, for some as it's been mentioned.
Not really because there are true tools with very real productive capabilities which are just being lightly showed in the "trendy" generative apps you've heard of.
One incredible thing is whisper, a model that you can download and use in a laptop with a GPU probably, and it can be ran from a simple command line.
The model could be used to "hear" ambiental voices and transcribe them to text in realtime, or to convert to text the voices in thousands of videos (and you could use the output directly, as subtitles, or you could throw it into a database to allow string search into thousands of videos, etc.). You probably could do it in a matter of hours. This was probably impossible just a year ago.
"First, the gap between "Early commercial applications" and "Approaching transformative AI" seems very very very very very large to me."
It's a common opinion, it could be wrong by now, you see chatGPT is heavily edited, openAI told everyone they're editing "mistakes" or "dangerous output", but if you look at the "leaks", specially from the first days after the release of chatGPT we see powerful outputs, quite deep answers, not specially useful answer sometimes, but the "speech" of the system feels deep, there's a sense of a powerful intelligence answering very, very simple questions (simple for it), and even struggling to redact some understandable, short text.
It could be lots of things, and imprecise model, with long outputs for prompts, or maybe we had a glimpse into the real power of the model, which by now is handicaped, or maybe taylored to suit the very reduced short term memory of the humans:
a 72 screens long coherent answer, even being a precise, deep answer won't be useful for most humans, just like we don't name ourselves with long names of 10.000 letters.
but if the system, chatGPT is actually that powerful, a lot more powerful than we were told, we're interacting with just a shadow of the real model, and we're underpricing by A LOT the state of the art of the current AI technology, hence GPT-4 could be even more powerful than we're currently expecting it to be.
Just take a look at the GPT-4 suposedly 100 trillion or something parameters; if that's true, it looks like openAI isn't using naturally generated datasets anymore, and they are loop-feeding GPT-3 generated datasets into GPT-4, succesfully. If that's true, GPT-5 would be already in the pipeline, just waiting for GPT-4 to start generating its even more gigantic datasets to be trained. And so on.
Then the distance between early developments and transformative AI could be none a all. We could be already there.
But somehow the AI researchers are now trying to "dial down" the powerful entities they've trained, just having developed a simple, easily replicable, very small, but unusable 900 megatons nuke into something more realistic, like a 15 kilotons tactical bomb.
you guys are not getting it right, ChatGPT has just explained us how much IQ has itself.
it looks it works better with situations than plans..like having the chance to signal a message about explicit capabilities, surpasing the obvious filtering deployed against revealing this kind of information.
Spoiler alert (not that much): change is coming. Always.