Olivia Vespera
Khen Ofek@OliviaVespera @amici @tinker
#AI is a general term which spans decades of development and huge amount of people.
In the 80' and 90' we used to talk about "expert systems" as the leading AI systems.
In the beginning of the 21st century we mainly talked about #machinelearning , and now we talk about LLMs and Multi-Modal Models.
The AI age is a loose term which looks at these developments from 50,000 feet.
I did study my history of AI. I'm familiar with the rule based and expert systems as well as the crash in AI.
So are you talking about the AI age as in the ones from the 80s onwards?
I'd say those old rule based expert systems are constructed meticulously and in a different way to the ones currently built that are based on the perceptrons.
I'm skeptical of current AI actually being able to do the work it's said to be able to: https://worksinprogress.news/p/why-ai-isnt-replacing-radiologists
@OliviaVespera @amici @tinker
I am talking about #AI age as a general abstract term.
I don't see why I need to go into the detail of the algorithm in the context of a philosophical economic discussion.
But, if you observe current #generativeai systems, you can find that they are very powerful without humans needing to construct the algorithm themselves, just a meta algorithm which learns by ltself.
I don't know what will become of #generativeai , but it looks very interesting.
@khenofek @amici @tinker Could you explain a bit more about what you're referring to when you say it's a general abstract term. I'm having trouble understanding what you're referring to.
As linked, genAI's aren't all that folks have been trying to sell them as. They also hallucinate and attack the work being done in open source software.
their economic promise is still debatable.
It's important to note that while they seemingly can do things, they don't actually understand what they are doing.
@OliviaVespera @amici @tinker
#generativeai in itself is also a broad topic and it is changing in a breakneck pace. We can talk all day long about new research and new tools that are being developed.
To talk about one ideology and one set of assumptions and purposes which underlies all of #AI and #generativeai seems nonsense to me.
#generativeai in itself is also a broad topic and it is changing in a breakneck pace. We can talk all day long about new research and new tools that are being developed.
To talk about one ideology and one set of assumptions and purposes which underlies all of #AI and #generativeai seems nonsense to me.
the nature of deep neural nets. The generative AI that we're talking about today an these new tools you read about are based on deep neural networks.
That's not going to change anytime soon. So the black box assumption is quite a fair assumption to make on GenAI, because it is the issue with the underlying technology.
@OliviaVespera @amici @tinker
Those deep neural nets are the current leading paradigm, although models like Random Forest are also used for specific use cases.
You are right that the black box issue is a real problematic one, but it is not an assumption.
Many people are trying to open that black box and pick inside, and they are having some progress.
Those deep neural nets are the current leading paradigm, although models like Random Forest are also used for specific use cases.
You are right that the black box issue is a real problematic one, but it is not an assumption.
Many people are trying to open that black box and pick inside, and they are having some progress.