zoogeny

17 ms·

181.
by zoogeny 2y ago
I think trial-and-error hand-waving isn't all that far from experimentation. As an aside, I was working in the games industry when multi-core was brand new. Maybe Xbox-360 and PS3? I'm hazy on the exact consoles but there was one
182.
by zoogeny 2y ago
This is a larger idea, just tangentially related to this particular case. In 2011 there was Occupy Wall Street. It was a movement that argued that many of the financial problems we saw in 2008 were a result of a 1% of wealthy business peopl
183.
by zoogeny 2y ago
Not directly related, but one of those stories that is so bizarre you almost can't believe it isn't made up. There was a NASA funded attempt to communicate with Dolphins. This eccentric scientist created a house that was half wate
184.
by zoogeny 2y ago
Interestingly, my criticism is exactly the opposite of yours. I think as LLMs become more and more capable (and crucially multi-modal) we will need the external tools less and less. For example, why would I want an MCP that can drive Photos
185.
by zoogeny 2y ago
I've heard the opposite, stories from users that MCP has been a flop. One of my acquaintances works in marketing and is a heavy LLM user. He got swept up in the hype of MCPs and tried a bunch of them out. He was expressing his disappoi
186.
by zoogeny 2y ago
Companies that want to maintain a data moat will work to maintain that moat in the face of any new technology. But not all data in the world is protected in that way. The use cases they promote are just easy to grasp, although they are misl
187.
by zoogeny 2y ago
I think we are confusing the word "prompt" here leading to miscommunication. There is the prompt that I, as a user, send to OpenAI which then gets used. There there is "prompt" which is being sent to the LLM. I don'
188.
by zoogeny 2y ago
Yes, but then you have to add that yourself to every prompt. It would be nice to tell your LLM provider just once "here is a tool you can use" along with a description of the API documentation so that you could use it in a bunch o
189.
by zoogeny 2y ago
Sure, at first, until the users demand a "always allow this ..." kind of prompt and we are back in the same place. A lot of these issues seem trivial when we consider having a dozen agents running on tens of thousands of tokens of
190.
by zoogeny 2y ago
> But that overhead is gonna be fixed-cost, and paid one-time at e.g. session initialization, not every time per e.g. request/response. I don't work for a foundational model provider, but how do you think the tool definitions g
191.
by zoogeny 2y ago
I don't think that could be sufficient to solve the problem. I'm using Gemini with AI Studio and the size of a 1 million token context window is becoming apparent to me. I have a large conversation, multiple paragraphs of text on
192.
by zoogeny 2y ago
I was unfortunate enough to work with SOAP and WSDL. There was a pipedream at the time of automatically configuring services based on WSDL but it never materialized. What it was very good at (and still has no equal to my mind) was allowing
193.
by zoogeny 2y ago
The bias of the article is strong, and I appreciate that. I actually despise performative unbiased reviews. While that bias initially frustrated me, once again comparing some contemporary thing a writer disagrees with to the rise of the Naz
194.
by zoogeny 2y ago
I still remember decades and decades ago hearing about vending machines in Japan. Someone mentioned going to Japan and how you could get cold cans of coffee out of vending machines there. This was sometime in the 1990s, before even Starbuck
195.
by zoogeny 2y ago
I've been thinking about this for a while. Seeing paid services like Focusmate, Flow Club and Focus101 show the basic idea is attractive enough to make into a lifestyle sized business. But my idea was a bit more. A lot of people are le
196.
by zoogeny 2y ago
This is one of the most promising uses of LLMs that I have found in my own work. Many times I have an idea for a refactor or even a feature but I have this mental reluctance just due to the amount of code I would have to write. Like, I have
197.
by zoogeny 2y ago
I want to make sure I am understanding this. If I have an application that uses OpenAI models then this service can act as a proxy between the my application and the actual OpenAI service. It logs all of the requests that get sent to the Op
198.
by zoogeny 2y ago
> An alternative for saving time with recompilation was to update my script so that only new markdown files or changed ones are marked for recompilation. That would involve too much wizardry if I wanted to make the script nice and robust
199.
by zoogeny 2y ago
I also haven't used the $200 exclusive model. But I think Gemini 2.5 is by far and away the best LLM I have used. I still use Claude for code but for any general reasoning questions, or even just for Socratic dialogue purposes, Gemini
200.
by zoogeny 2y ago
I just want to jump on this because I was thinking about it just yesterday. I was raised Catholic but I am not religious and based on the creeds I would not qualify as a "true" Christian since I reject miracles like virgin births
201.
by zoogeny 2y ago
I think the trend of "Spiritual But Not Religious (SBNR)" is the most interesting. I suspect a large number of people leaving religions aren't militant atheists convinced by the logic of Dawkins/Hitchens/Harris et a
202.
by zoogeny 2y ago
I was confused when xAI was started as a separate company. Twitter had the data, the delivery surface, the user base. I wondered why a separate entity was created to house the AI part. I should have had a clue because I was talking to a CEO
203.
by zoogeny 2y ago
When you say "those people", you mean people like me who (used to) have to navigate how to model structured data using XML. I think the attribute vs. child distinction makes sense in a very flat hierarchy where you are marking up
204.
by zoogeny 2y ago
I see, so my example should be: (foo (bar "baz") "content") vs (foo ((bar "baz")) "content") Where the first one would be the nested tags and the second one would be a single `bar=&q
205.
by zoogeny 2y ago
This looks like it loses the distinction between attributes and nested tags? As in, I don't see a difference between `(attr "val")` which expresses an attribute key/value pair and `(thing "world")` which expres
206.
by zoogeny 2y ago
Yes, and one of my points was: if metal becomes as important to musical history (over a similar time span) as classical romantic music or jazz it will get the same treatment. I'm arguing that reasoning "classical music is importan
207.
by zoogeny 2y ago
I don't see much difference in that compared to now. I mean, do a search on Jazz composition/improvisation and you are bound to find numerous descriptions of the Barry Harris approach (among others). There is a large existing cann
208.
by zoogeny 2y ago
What I'm getting at is that the literate quality of the music as a defining principle (or differentiator) is lessened if the actual practice was improvisation that preceded it being written down. If a composer improvised the music fi
209.
by zoogeny 2y ago
I didn't study classical, but I was watching some YouTube channel by someone with graduate degree in classical. He offhand mentioned that many of the "compositions" of guys like Chopin were actually improvisations that were l
210.
by zoogeny 2y ago
I think the "literally cry" thing is a zoomer thing. I watch a lot of YouTube music video channels and the "this made me cry" thing is a meme that shows up with extreme frequency. I don't think we should take it any

More ›