larryfreeman

One week ago, Geoffrey Hinton posted on X about the Loch Ness Monster. I thought that there would be a follow up with more explanation. I am unclear on his intent. Was this an actual sighting, an explanation of an illusion which appears like the Loch Ness Monster, did someone hack his account, or was Hinton posting under the influence: https://x.com/geoffreyhinton/status/1729247583061553361?s=20

As a non-expert, ChatGPT appears to me to be a compression of the internet with interesting mash ups. Am I wrong? Creativity to me is about going outside a given search space, about disruptive, non-linear change. ChatGPT is about a constrained search space that is unbelievably large. I would love to hear the reasoning for why my thinking is incorrect so I can better understand the creative potential and limitations of ChatGPT.

Fuzzy Logic is a logic based on an imprecise value for true and false. It seems to me that from a perspective of imprecise concepts and conditions, LLMs allow the same feature for programming: Instead of stating the exact method of measurement and the exact threshold, we can now describe a condition intuitively in plain language and rely on an LLM such as ChatGPT to determine if the condition applies to the current context. We can thus do "fuzzy programming" with algorithms based on imprecise categories, conditions, and entities. Are there any technical papers or blogs that folks can recommend that talk about the unlocked capabilities of fuzzy programming? I am especially interested in design patterns and any discussions on features that are greatly improved by imprecise programming.

I'm not too active on technology forums. I think that Hacker News and Stack Exchange are my only places where I am somewhat active.

As a result, it is quite challenging to write up and present a naive idea to find out if it has any meat. In general, expressing an idea without being clear on the technical history or misusing a technical term is an automatic disqualification of the idea independent of the value of the idea.

I am curious. Is there a recommended approach that more successful folks take when you have a potential solution to a well-known problem that the experts believe cannot be solved by a non-expert.

In my case, I've been thinking a lot of about alternatives to using a private key as the basis of identify for decentralized currencies. I have come up with an approach (that is most likely flawed and may not have any meat) for a decentralized identity service that is not based on private keys (in essence: if the private key is lost, a new one can be used in its place).

What would be the recommended approach for finding mentors, presenting the early draft, and ideally, receiving constructive feedback that will help me understand how to place the approach that I came up with within the context of current efforts.