4 ms·

I'm not sure this is Ben's forte, but you're right that I wouldn't be surprised if he tries it, though he has done some circuit stuff[0,1] so nothing would surp

I'm not sure this is Ben's forte, but you're right that I wouldn't be surprised if he tries it, though he has done some circuit stuff[0,1] so nothing would surprise me from him. (Hi Ben! Love the work!) BUT I do think this is something Sam Zeloof[2] try. He's done some lithography using a projector[3]. Also there's Jeri Ellsworth, but I think she's shifted to mostly working with her AR project. Tons of old videos on that stuff if you're into it. Side note: I'm assuming anyone who knows any of these people would be interested that a new Dan Gelbart video just dropped[5]! ----------------------------------------- Other side note: @YouTube people (and @GoogleSearch), can we talk about search? The updates have been progressively making it harder to find these types of accounts. People who do *highly* technical things. I get that these are aimed at extremely specific audiences but this type of information is some of the most valuable information on the planet. Lots of knowledge is locked into people's heads and these classes of videos are one of the biggest booms to knowledge distribution we've ever seen in the course of humanity. I understand that this does not directly lead to profits to YouTube (certainly it DOES for Google Search), but indirectly it does (keeps these users on your platform!) and has a high benefit to humanity in general. The beauty of YouTube and Google was you could access anything. That we recognized everyone was different and we could find like minded people in a vast sea. The problem search was meant to solve was to get us access to hard to find things. To find needles in ever growing haystacks! Please, I really do not want to return to the days of pre-search. Nor even early search! It should be _easier_ to find niche topics these days, not harder. LLMs aren't going to fix this. This is becoming an existential crisis and it needs to be resolved. [0] https://www.youtube.com/watch?v=UIqhpxul_og https://www.youtube.com/watch?v=UIqhpxul_og [1] https://www.youtube.com/watch?v=FYgIuc-VqHE https://www.youtube.com/watch?v=FYgIuc-VqHE [2] https://www.youtube.com/@SamZeloof https://www.youtube.com/@SamZeloof [3] https://www.youtube.com/watch?v=XVoldtNpIzI https://www.youtube.com/watch?v=XVoldtNpIzI [4] https://www.youtube.com/@JeriEllsworthJabber https://www.youtube.com/@JeriEllsworthJabber [5] https://www.youtube.com/watch?v=OuZjjActWmQ https://www.youtube.com/watch?v=OuZjjActWmQ

PaulHoule 2y ago
LLMs could help if they were specifically applied to the task [1], however people are actually applying them to the generation of countless slop videos. Google's problem, which I think there is no cure for, is that Google believes it is #1 and to quote Fatboy Slim "We're #1 why try harder?" If in some way they feel they have competition it is to be a 2nd rate TikTok, not be a better version of what made YouTube great. In the meantime, for everybody that's been turned on to something really awesome and creative on YouTube somebody else got turned on to something really toxic. [1] Something significant happens every 10 years in search relevance, and SBERT was one of those.
- ezst 2y ago
  Again someone mistaking LLMs with knowledge bases. Must be a day finishing in `y`
  - PaulHoule 2y ago
    The original misunderstanding behind "knowledge base" was that, in the 1980s, it was an idea in symbolic AI that you'd develop a set of facts against an ontology designed for accurate inference and somehow by the 1990s it became a text repository with a search engine that may or may not work. Occasionally useful, sometimes hard to distinguish from a trash can. See Confluence. Prompt engineers with their decoder models are going to always be wondering why they are always a bridesmaid and never a bride, with encoder models you can attain the holy grail of the system where you put text in one side and get, within calibrated accuracy, facts to put into the first kind of knowledge base. Or, for that matter, a good search engine for the second kind of knowledge base which could raise it above the "trash can" level.
    - ezst 2y ago
      "Funny" how that reminisces of the whole blockchain discussion. If the need is fully satisfied by a "boring" and cost-effective "facts" database, why would an adequate engineer push for (blockchain/)LLM instead?
      - PaulHoule 2y ago
        There were several reasons why "expert system" were rejected in the 1980s including competition with programmable calculators and spreadsheets and no correct paradigm for reasoning with uncertainty but the one most quoted was that the creation of that kind of database is not cost-effective. I spent about 10 years working (sometimes for myself, sometimes for employers, sometimes part time, sometimes as a software developer sometimes as a business developer) on the problem of turning a mass of text into facts into text to solve problems like: - Doctors write copious medical notes from which facts would be useful for themselves, payers, researchers, regulators. - An accounting or legal firm may need to scan vast numbers of documents and extract facts for a audit or lawsuit - An aerospace manufacturer has a vast database of documentation and maintenance notes (even from the teams at the airports) that it needs to keep on top of - A fashion retailer wants to keep track of social media chatter to understand how it connects and fails to connect with customers and answer questions like "should we endorse sports star A or B?" - Police and soldiers chat with each other over XMPP chat about encounters with "the other" which again are rich with entities, attributes, events, etc. Tasks like this need an interactive system but you face the problem that people have an upper limit of 2000 or so simple decisions [1] in a sustainable day. The problem is large but it is not "boil the ocean" because you can set requirements for what gets extracted and use the techniques of statistical quality control as in Deming to know accuracy is in bounds. You can give people tools to tag things in bulk, you can apply rules, you can give the people tools to create the rules. I worked on RNN and CNN based models, SVM, logistic, autoencoder and other models and before BERT they all sucked. If you have the interactive framework you can put encoder or decoder LLMs in and it is a revolution that makes systems like that much cheaper to develop and run for better effects. [1] hot dog/not hot dog

PaulHoule 2y ago

LLMs could help if they were specifically applied to the task [1], however people are actually applying them to the generation of countless slop videos. Google's problem, which I think there is no cure for, is that Google believes it is #1 and to quote Fatboy Slim "We're #1 why try harder?" If in some way they feel they have competition it is to be a 2nd rate TikTok, not be a better version of what made YouTube great. In the meantime, for everybody that's been turned on to something really awesome and creative on YouTube somebody else got turned on to something really toxic. [1] Something significant happens every 10 years in search relevance, and SBERT was one of those.

ezst 2y ago
Again someone mistaking LLMs with knowledge bases. Must be a day finishing in `y`
- PaulHoule 2y ago
  The original misunderstanding behind "knowledge base" was that, in the 1980s, it was an idea in symbolic AI that you'd develop a set of facts against an ontology designed for accurate inference and somehow by the 1990s it became a text repository with a search engine that may or may not work. Occasionally useful, sometimes hard to distinguish from a trash can. See Confluence. Prompt engineers with their decoder models are going to always be wondering why they are always a bridesmaid and never a bride, with encoder models you can attain the holy grail of the system where you put text in one side and get, within calibrated accuracy, facts to put into the first kind of knowledge base. Or, for that matter, a good search engine for the second kind of knowledge base which could raise it above the "trash can" level.
  - ezst 2y ago
    "Funny" how that reminisces of the whole blockchain discussion. If the need is fully satisfied by a "boring" and cost-effective "facts" database, why would an adequate engineer push for (blockchain/)LLM instead?
    - PaulHoule 2y ago
      There were several reasons why "expert system" were rejected in the 1980s including competition with programmable calculators and spreadsheets and no correct paradigm for reasoning with uncertainty but the one most quoted was that the creation of that kind of database is not cost-effective. I spent about 10 years working (sometimes for myself, sometimes for employers, sometimes part time, sometimes as a software developer sometimes as a business developer) on the problem of turning a mass of text into facts into text to solve problems like: - Doctors write copious medical notes from which facts would be useful for themselves, payers, researchers, regulators. - An accounting or legal firm may need to scan vast numbers of documents and extract facts for a audit or lawsuit - An aerospace manufacturer has a vast database of documentation and maintenance notes (even from the teams at the airports) that it needs to keep on top of - A fashion retailer wants to keep track of social media chatter to understand how it connects and fails to connect with customers and answer questions like "should we endorse sports star A or B?" - Police and soldiers chat with each other over XMPP chat about encounters with "the other" which again are rich with entities, attributes, events, etc. Tasks like this need an interactive system but you face the problem that people have an upper limit of 2000 or so simple decisions [1] in a sustainable day. The problem is large but it is not "boil the ocean" because you can set requirements for what gets extracted and use the techniques of statistical quality control as in Deming to know accuracy is in bounds. You can give people tools to tag things in bulk, you can apply rules, you can give the people tools to create the rules. I worked on RNN and CNN based models, SVM, logistic, autoencoder and other models and before BERT they all sucked. If you have the interactive framework you can put encoder or decoder LLMs in and it is a revolution that makes systems like that much cheaper to develop and run for better effects. [1] hot dog/not hot dog