I hear world-melting tech rumbling our way

Large Language Models (LLMs) are tingling my antennae a lot -- the
way the Alto and the Web did. But I think this evolution is going
faster. Without getting into too many details I don't understand,
the large language part of it is that the models are based on very
large collections of texts, images, sounds, code. (I wrote here about using LLMs to write back in May, 2021) So if it weren't
for all of us putting everything online over the past three decades,
there wouldn't be anything to apply machine learning to. But the way
significant knowledge can be parsed out of it is, IMO, impressive
although the technology is in its infancy. Yes, it swallowed all the
bullshit along with the good info, and yes, it is unreliable and
makes up shit, and no the models are tools. They don't understand.
They do statistics. Now Google is beta-testing Bard, a chat
interface. And Bing (Microsoft has $10 billion in the company that
made chatGPT) has already incorporated chatGPT.  I searched for
something that isn't as easy to find via Google: What is the
difference between coordination, cooperation, and collaboration, and
I got this result:

A couple days ago I got wind of a tool that can unprocess an image
generated by midjourney, dall-e, et. al. to surface the images it
was based on.

I see radical changes ahead. And if we are honestly looking back at
the last decades of rapid technological change, we're in for a world
of hurt along with some real miracles in fields like diagnosis and treatment of disease and creating effective medications for emerging diseases rapidly.

For example I sense that we are on the cusp of an inflection point
in the conduct of science as significant as the introduction of
computers: the use of  generative AI as scientific thinking and
knowledge tools.

Just one example that is already being replicated in parallel fields is the computational biology of proteins.

Proteins are topologically complex & can fold into a  large number
of possible shapes. Much of  immune system & anti-cancer therapies
rely on matching the shape of proteins on the surface of a cell.
Now, AI can propose  (previously unknown) proteins of medical
significance.


Wanna get scary about that? Generative AI suggested 40,000 chemical
weapons in just six hours


I read someone in the #technology stream on Mastodon proposing that
a good way to think about a proposed technology is to ask: What
would 4chan do with it?

Connecting computational biology to wetlab synthesizers is just a
matter of money and expertise. What will 4chan, ISIS, or the Boogaloo Boys do with LLM tools?