visarga

27 ms·

541.
by visarga 1y ago
> The Dotcom boom was probably good for everyone in some way, but it was much, much better for the extremely wealthy people that have gained control of everything. You are describing platform capture. Be it Google Search, YouTube, TikTok
542.
by visarga 1y ago
> It will benefit all people but it will disproportionally benefit more rich people The rich already have a diminishing returns situation with money. Everyone else has much more upswing.
543.
by visarga 1y ago
Richard makes a distinction between human understanding and AI indifference to truth. But isn't that what half the country is doing a.t.m? And more philosophically, we can't know the Truth because we rely on leaky abstractions all
544.
by visarga 1y ago
> having access to precursors is just as important as having access to the weights They probably can't give you the training set as it would amount to publication of infringing content. Where would you store it, and what would you d
545.
by visarga 1y ago
I saw on X someone posted a screenshot where Gemini got depressed after repeated failure, apologized and actually uninstalled itself. Honorable seppuku.
546.
by visarga 1y ago
Yeah, I had Cursor refactor a large TypeScript file today and it used a script to do it. I was impressed.
547.
by visarga 1y ago
Cursor likes to create one-off scripts, yesterday it filled a folder with 10 of them until it figured out a bug. All the while I was thinking - will it remember to delete the scripts or is it going to spam me like that?
548.
by visarga 1y ago
I think the most important insight from your article, which I also felt, is that agentic search is really different. The ability to retarget a search iteratively fixes both the issues of RAG and grep approaches - they don't need to be
549.
by visarga 1y ago
> once AI is capable enough to do all tasks humans do in employment, there will be no need to employ any humans at all for any task whatsoever AI has no skin, you can't shame it, fire it, jail it. In all critical tasks, where we tak
550.
by visarga 1y ago
I tried Claude Code once and half an hour later it printed $10 cost. I thought I was using the pro subscription, not the API. This makes using CC dangerous, so I am avoiding it.
551.
by visarga 1y ago
> (I work at Cursor) I find the amount of credits included in the pro subscription per month totally insufficient. Maybe it lasts 1-2 weeks. Today I got a message telling me I exhausted my subscription when the web dashboard was showing
552.
by visarga 1y ago
What I hate about Cursor is that - even when I have credits left - it still hangs for a full minute before starting to respond. Not always, but often enough. You don't know what you're buying, you might get fast response, you migh
553.
by visarga 1y ago
> the ability to include video clips of people and products as a part of the prompt and then create a realistic video with that This is something I would not like to see, I prefer product videos to be real, I am taking a risk with my mon
554.
by visarga 1y ago
> Claude Code, Codex CLI etc can effectively do anything that a human could do by typing commands into a computer. They still don't have good integration with the web browser, if you are debugging frontend you need to carry screensh
555.
by visarga 1y ago
> assumes that 10x coding speed should 10x productivity This same error in thinking happens in relation to AI agents too. Even if the agent is perfect (not really possible) but other links in the chain are slower, the overall speed of th
556.
by visarga 1y ago
Discovery comes from search for both humans and AI agents. There is no magic in the brain or LLM except learning along the way and persistence. The search space itself is externalized. So the AI agents are "good enough" but enviro
557.
by visarga 1y ago
> The child and the wolf try to enjoy a picnic by the river but there's a sheep. The boat needs to connect nine dots over the river without leaving the water but gets into an accident and dies. The surgeon says "I can't op
558.
by visarga 1y ago
I think you jump from AGI to "human not needed" too abruptly. First of all, AGI might be smarter than you but you have to live with the consequences of using it. So you can't be removed from the loop. We need accountability,
559.
by visarga 1y ago
Things still cost money. There are always scarcities. For example cells already reached exponential self replication capabilities long ago, but eventually hit environment constraints. It became a struggle for survival, but under infinite re
560.
by visarga 1y ago
Assuming that 1. food is free and instant to get, and 2. there are infinite possibilities for food - then yes, if you ordered such a food from an infinite catalog you would get the credit. But if you ordered 100 dishes iterating between des
561.
by visarga 1y ago
Maybe you need to refresh the context - 99.99% of AI generated music, images or text is seen/heard only Once, by the AI user. It's a private affair. The rest of the world are not invited. If I write a song about my kid and cat it&
562.
by visarga 1y ago
You probably never prompted an AI model until you got something good from it.
563.
by visarga 1y ago
What button? Again with the vending machine idea. No, it's language prompting, language has unbounded semantic space. It's not one choice out of 20, it's one choice out of countless possibilities. I give my idea to the model,
564.
by visarga 1y ago
I like your song.
565.
by visarga 1y ago
I wanted to try the advanced Suno models but they are locked away, and I don't want to take a membership just to try a couple of generations. So, for months, I gave up on Suno. It's locked away at v3.5 for me. Last 3 models have b
566.
by visarga 1y ago
> My 8 year old nephew was cranking out "songs" in Suno within an hour of being introduced to it. Reminds me of when parents were super impressed that their 3-year old could use an iPad. You imply "it is Prompt -> Song&
567.
by visarga 1y ago
I would use Deep Research mode outputs. Sometimes I run multiple of these in parallel on different models, then compare between them to catch hallucinations. If I wanted to publish that, I would also doublecheck each citation link. I think
568.
by visarga 1y ago
> But more than that, I think, it shows that there was no magic to AlphaFold. The details of the architecture and training method didn't matter much. All that was needed was training a big enough model on a large enough dataset. Peo
569.
by visarga 1y ago
Or you plug it into one of your old monitors. We all have them laying around.
570.
by visarga 1y ago
For me the most annoying text selection limitation is on iOS - there is no "Select All". I have to scroll for a full minute to get a page.

More ›