“Most of this Tech is still in the conceptual stage & it doesn't have a clearly defined scope / boundary.”
LOL
Man I’ve been asking people what is this AGI and nobody can give me a non hand wavy explanation.
And I’m asking this question as someone who has been running AI workloads since 2010 and can write a tiny neural net to recognize characters from scratch.
You are the first I’ve seen to say basically they don’t know what AGI is.
But we don’t know how the NNs work up to a point either, so training this thing is trial and error to find what tweaks works best with what problem space.
Another good article is “how does ChatGPT work” by Stephen Wolfram.
Yes Shawn, I agree. The pace at which Tech is evolving is mesmerizing to say in the least.
Most of AGI is still conceptual today, but is evolving constantly & at a decent pace as well. And when you talk of NN you're essentially referring to DL which is to say "a drop in the ocean" given today's parlance of AI (ANI actually).
Compared to my first stint with AI back in 2012-13 where we built an analytics engine that could use up a variety of first-hand market data & metadata to come up with a high-level view of an asset, I think this Tech has seen some handsome progress.
Given the spate of AGI today, I have no doubt that it has got pretty advanced, but is still being subject to a lot of testing given the necessity to tick off the "feasibility" & "viability" so that it could be labeled "totally foolproof".
Thanks for that reference in the end, shall surely check that out.
Re AGI
“Most of this Tech is still in the conceptual stage & it doesn't have a clearly defined scope / boundary.”
LOL
Man I’ve been asking people what is this AGI and nobody can give me a non hand wavy explanation.
And I’m asking this question as someone who has been running AI workloads since 2010 and can write a tiny neural net to recognize characters from scratch.
You are the first I’ve seen to say basically they don’t know what AGI is.
But we don’t know how the NNs work up to a point either, so training this thing is trial and error to find what tweaks works best with what problem space.
Another good article is “how does ChatGPT work” by Stephen Wolfram.
Yes Shawn, I agree. The pace at which Tech is evolving is mesmerizing to say in the least.
Most of AGI is still conceptual today, but is evolving constantly & at a decent pace as well. And when you talk of NN you're essentially referring to DL which is to say "a drop in the ocean" given today's parlance of AI (ANI actually).
Compared to my first stint with AI back in 2012-13 where we built an analytics engine that could use up a variety of first-hand market data & metadata to come up with a high-level view of an asset, I think this Tech has seen some handsome progress.
Given the spate of AGI today, I have no doubt that it has got pretty advanced, but is still being subject to a lot of testing given the necessity to tick off the "feasibility" & "viability" so that it could be labeled "totally foolproof".
Thanks for that reference in the end, shall surely check that out.