If they conceptualize, why do they sometimes spit out nonsensical BS?
Let’s flip this around - How can you tell the difference between an LLM being able to conceptualize yet being wrong sometimes vs. not being able to conceptualize?
Without knowing anything about machine learning and bearing in mind AI is super hyped up with marketing BS right now, it sounds like “emergent properties” are in the eye of the beholder and not actually evidence of some higher order intelligence at work.
We already had this, it’s called Intel Optane Persistent Memory and Intel killed it off last year: https://www.intel.com/content/www/us/en/products/docs/memory-storage/optane-persistent-memory/overview.html
The memory speed was slightly slower than DDR4 but the benefits didn’t seem to outweigh the downsides. I think it probably kicked a lot of ass for specific use cases (eg. in-memory database that needs persistence), but the market was too small. Plus, SSDs are getting so ridiculously fast that it would put pressure on a product like this too.
IMHO all these approaches are convoluted and introduce way too many components (SPOFs) to solve the problem. They’re “free” but they come at the cost of maintaining all this extra infrastructure and don’t forget that certificate transparency logs mean all your internal DNS records that you request a LetsEncrypt certificate for will be published publicly. (!)
An alternative approach is to set up your own internal certificate authority (CA), which you can do in a couple minutes with step-ca. You then just deploy your CA root cert to all the machines on your network and can get certs whenever you need. If you want to go the extra mile and set up automatic renewal, you can do that too, but it’s overkill for internal use IMHO.
Using your own CA introduces only a single new software component and it doesn’t require high availability to be useful…
There’s a business strategy called embrace, extend, extinguish that they’ll try to use to snuff out the fediverse.
This was really insightful, thank you! I also loved how Bard’s output completely mistakes the common physics riddle. (I have a physics background and your analysis is spot on IMHO.)