tkgally

52 ms·

571.
by tkgally 3y ago
One of the students in a course I’m teaching on language and AI (mentioned in another comment here) wrote something similar in a homework assignment the other day. We had discussed the paper “Consciousness in Artificial Intelligence: Insigh
572.
by tkgally 3y ago
As it happens, I’m now teaching an undergraduate course at the University of Tokyo titled “The Meaning of Language in the Age of AI.” The students—who come from various countries and majors—and I are discussing how theories of human languag
573.
by tkgally 3y ago
There's an interesting contrast between that slick faked demo with the seemingly live video recognition and another video Google released the same day, showing the processing and understanding of raw audio [1]. In the latter, the awkwa
574.
by tkgally 3y ago
Just out of curiosity: Are you hiring only experienced developers? Or are you also hiring junior developers who are just getting started in the field? Discussions here on HN about whether or not AI is a threat to developers’ jobs seem to su
575.
by tkgally 3y ago
Mine as well—and not only paper but writing implements as well. I don’t use pens or pencils much anymore, but I still get a kick out of browsing and trying out the vast selection in large stationery stores in Japan. (I live in Yokohama.) Ja
576.
by tkgally 3y ago
> These kind of live translations while impressive require the full pipeline to work, speech recognition and then proper translation. That's a good point. Machine translation of text input by LLMs is pretty good for the languages I
577.
by tkgally 3y ago
Interesting paper. I just tried it myself. I first had GPT-4 write a Python program to randomly reorder the letters in each word in three sentences I took from the New York Times website. I then started a new GPT-4 session and used the same
578.
by tkgally 3y ago
I get "intriguing" a lot from GPT-4. As well as replies that conclude with "In conclusion,...."
579.
by tkgally 3y ago
Excellent! Here is my first attempt: Me: How can I write a good essay? MonadGPT: The best way to write a good Essay is by following these instructions: First of all, you must have a good and well-grounded knowledge in that subject which you
580.
by tkgally 3y ago
Yes, you need to be aware of the possibility of confabulation by the AI. But you can minimize that problem by: - Using an LLM that hallucinates less. GPT-4 is more reliable than GPT-3.5, and the just-released Claude 2.1 is reportedly better
581.
by tkgally 3y ago
> It is very trustable about high level things many people know. As the knowledge is more specific or esoteric the less trustworthy it is. That's my experience, too. In addition to sticking to general topics, another fun way to talk
582.
by tkgally 3y ago
> I’ll just pick a topic and then ask for a summary and then ask clarifying questions. I made a public GPT for a similar purpose. Some examples of my conversations with it: https://youtube.com/@Tom_Gally_UTokyo
583.
by tkgally 3y ago
In addition to commercialization providing money for AI development, isn't there also the argument that prudent commercialization is the best way to test the models for possible dangers? I think I saw Mira Murati take that position in
584.
by tkgally 3y ago
Thank you for that comment. There are many people here who seem to believe that the OpenAI employees--and nearly everyone else involved in this drama--are motivated only by money. While of course money matters, people are also motivated by
585.
by tkgally 3y ago
> It's only used by newscasters, and car dealers in their commercials. When I was growing up in Southern California, “the Southland” was used in the same way to mean the Los Angeles area. I heard the term only in the mass media, nev
586.
by tkgally 3y ago
The Moon Is A Harsh Mistress by Robert Heinlein has as one of its central plot elements a mainframe computer that becomes sentient and able to converse with humans. It’s been more than fifty years since I last read the book, but it has re
587.
by tkgally 3y ago
> This is really good journalism. That was exactly my reaction. I’ve been following the news and rumors and speculation closely since Altman’s firing, and this is by far the most substantive account I have read. Kudos to the authors and
588.
by tkgally 3y ago
I didn’t have much sense of who Ilya Sutskever is or what he thinks, so I searched for a recent interview. Here’s one from the No Priors podcast two weeks ago: https://www.youtube.com/watch?v=Ft0gTO2K85A No clear clues abou
589.
by tkgally 3y ago
For me, the most valuable "take" on generative AI has been the collective discussion on HN. While many individual articles and comments have been useful on their own, it is the ongoing conversation here that has given me the best
590.
by tkgally 3y ago
Well, in my case, I was struggling to come up with a coherent understanding for myself about how views of language might need to change when interaction with LLMs is taken into account. ChatGPT’s response helped me think about the issue mor
591.
by tkgally 3y ago
Curious about what ChatGPT might have to say on this issue, I asked it: https://www.gally.net/temp/20231116languageandai/index.html
592.
by tkgally 3y ago
The article focuses on linguistic theories that regard natural language in a somewhat narrow sense: language as a cognitive phenomenon, language as a symbolic system, language as something that can be modeled and imitated by a computer. Whi
593.
by tkgally 3y ago
As one gets deeper into learning another language, it’s also important to be aware that the meanings of words in different languages rarely map to each other in clean bijections. Common words especially tend to be semantic clouds, not fixed
594.
by tkgally 3y ago
That’s my experience, too. Most things work, but I get occasional errors and response is often much slower than a few days ago. Just now, I was editing a GPT and the text produced by the editor bot appeared at about human typing speed.
595.
by tkgally 3y ago
I submitted the link to this article before I read the article. Now I’ve read it, and I’m glad I did. It’s beautifully written, and it helps me understand better who Geoffrey Hinton is and why he is so worried about AI. The article is archi
596.
The Godfather of A.I. Fears What He's Built(newyorker.com)
6 points by tkgally 3y ago|1 comments
597.
by tkgally 3y ago
Yes, that is an important distinction. My guess is that the effective self-prompting is due not only to prompt engineering examples scraped from the web but also to some kind of automated reflexive training, in which the model produces mi
598.
by tkgally 3y ago
And just think how wonderful it is that that approach works. How many people would be able to tell you "This is how you should tell me what to do " both appropriately and without getting annoyed at the request?
599.
by tkgally 3y ago
> Music has never been in a better place. I have to agree. I’m 66 years old. When I was in my late teens, I was intensely into music and doing my best to explore and learn about a variety of genres. It was hard. I couldn’t afford to buy
600.
by tkgally 3y ago
Excellent! The first one I remember seeing—and enjoying—was “Prof or Hobo”: https://www.proforhobo.com/

More ›