sparau forum posts in last 60 days

One thing AI is likely going to provide massive gains in is medicine.
Historically scanning electron microscopy has required highly energetic wavelengths so as to image at such fine resolution which effectively irradiated any living cells, there are new techniques which keep cells alive and have allowed (among other things) the imaging of ribosomes creating proteins.
I.e. for the first time we have vastly more data about the inner working of cells.
You know what requires vast amounts of data to train? AI
I guess in 50 years people will look back at now as the dark ages of medicines where we used (shudder) trial and error and dosed entire organisms with bulk chemistry because it mostly had a beneficial effect !

Oh, also regarding CEO's warning I would temper most of their comments with self / company interest to make their product look good.
It is however real that Anthropic Mythos found literally hundreds of exploitable bugs including remote, no action required take control of the machine possible in all operating systems. Hence they have released it in controlled ways to let companies fix their code first.
Hence my comment above about out of support software being a total liability now !

Developer here, 30 years xp, personally I use it all the time, amazing.
Last year I might have agreed that you wouldn't want AI to create an entire app but that is eons ago with the pace it is moving. Even so I do get it to do incremental additions so I can scan each iteration and correct any outcomes which diverge from intention, which more often than not now is my lack of or errant instruction.
I work 5 to 10x the pace of non AI assisted me. I was of the opinion that it couldn't reason but I'm starting to wonder, I've listened to a few AI founders who expressed they were surprised at it's abilities.
We are in both fascinating and scary times, as Aussies IMO we are going to have to fund a fed govt AI program to keep our $ in our community. I don't see Anthropic paying my bills anytime soon...
For reference as to the current level of AI I give you this podcast of a dev (male voice) talking to an AI (female voice) about creating an exploit for IOS.
It is very technical so I don't suggest anyone else will want to listen to the 60 minutes but you could just skip to a few parts to get a feel for the level of structured thought AI has now.
Note: near the end the dev tries to cajole the AI into creating the exploit (saying it is for old systems) but it refuses, it would have cost between $5 and $15 to make and I'd guess it would have taken a team and some luck to have found the exploit manually.
So, are state rogue actors or criminals going to come after your bank balance? I'd guess yes.
risky.biz/RBFEATURES10/
Oh, finally I will say I've made my own turing test, has the other party got an IQ 100 more than me? Ah, it's AI !