Won't this be bad for agents since now you need to provide this new API in their prompt, as opposed to just regular git which the model has seen enough in its training data
history suggests so .... people do keep trying to make agent native tools and workflows, but time and time again it turns out to be better just to expose raw inputs and tools to them and let them work with those. See skills beating MCP in most cases where their purpose overlaps for example - it's more effective just to let an agent write git commands than give it a "git tool" with a structured interface. People don't seem to grok the intuition of how heavily biased training on trillions of token of human language and existing software code makes the models towards working well with raw input.
What is this based on? The only thing I can think of is AI coding tools but only a few companies do it properly. I don't see gitlab capturing any of that spending
Also the whole "removing layers". Today's prof g market video was about the topic. Afaik it was the Coinbase CEO telling the same. Do these people get together to discuss their talking points? Or are they signalling to investors?
Presumably based on the fact that the OpenAI/Anthropic $200/month plans are selling like hot-cakes, and it's not often that a new software category comes around which attracts those kinds of per-seat prices.
Why is every startup using that same Serif font now, Garamond or whatever. Is it an LLM design phenomenon? Its kinda ruining that font style for me.
Also $1,500 a month for 10 "influencers" is wild. This doesn't seem that sophisticated unless they're doing something special to increase trust scores of accounts. They say they have "in house warming algorithm" which honestly doesn't inspire confidence for me.
Whats funny is its almost a certainty (if they are doing things correctly) that they have literal farms of phones (probably in SEA). The only real way to keep trust high is to have a real mobile connection and unique devices. Proxies are okay, but you really need to use the apps on real hardware.
Interesting article, thanks. I've done a bit of small scale phone farming (for my own cheap mobile proxies). In all reality the phones aren't that expensive, I went with Moto 5gs that cost $130 (retail), so in their case the phones pay for themselves in the first month.
Probably a decent amount of compute cost for video generation, but I'm sure they have access to free compute and inference for being in bed with a16z.
How is this not grounds to be sued into oblivion by Google and Meta? They clearly violate ToS for profit. This is something I expect to find on a dark web forum where 0days are traded, not in public.
> How is this not grounds to be sued into oblivion by Google and Meta?
Because they don't care. It doesn't matter that it's AI slop, it generates views. And Google and Meta can bill advertisers for those views.
Zuckerberg is paying people to put AI slop Shrimp Jesus on facebook. (Not directly to platforms like this, but with the incentive structure)
Really, they're not just cashing in on the views of AI slop being put in front of boomers. They're cashing both ways; While the low end spam industry is merely guessing and iterating on whatever generates views, the more refined spammer does not leave the performance of their latest slop post up to chance, and just uses good old viewbotting. Viewbotting that these days, is mostly done on real devices. Which show ads, to the bots or underpaid developing world workers. Google and Meta'll still charge you for those impressions though.
The losers? People who sincerely try to use these platforms, and whatever idiot businesses are still paying for ads by the impression or click, rather than conversions that immediately generate revenue.
This kind of thing has been common for ages. Obviously AI has kicked it into overdrive, but it’s not darkweb kind of stuff.
Note that they do not mention any specific companies on that landing page. That is pretty intentional.
But realistically going after bots is expensive and rarely successful, so most companies don’t do it. Even if you find the guy, the chances they can be legally reached are pretty low.
I had to create a hypothesis testing agent where it gets a query like "is manufacturing parameter x significantly different this month than last month" and have the agent follow a flowchart to run a statistical test and return the answer
At the time I had access to only 4o and there was no way to guarantee that the agent would follow the flowchart if I just mention it in its prompt. What I ended up wrapping the agent in a loop that kept feeding it the next step in the flowchart. In a way, a custom harness for the agent
Not totally wrong. Self play works well with if your problem can be easily simulated in an RL environment where the model can easily explore different states. RLHF or similar techniques is not that since we don't have exactly have a simulation environment for language modelling
Right now there are companies which hire software devs or data scientists to just solve a bunch of random problems so that they can generate training data for an LLM model. Why would they be in business if self play can work out so well?
> Right now there are companies which hire software devs or data scientists to just solve a bunch of random problems so that they can generate training data for an LLM model.
> language itself is one tool of communication invented to supersede body-language and grunting and noises
That's a pretty utilitarian view of language. How would it feel if everyone spoke and wrote like a PR representative? This is what an article written by an LLM is starting to sound like.
I'm even willing to argue that the way in which you convey your ideas is as important as the idea itself. Like we could all be eating soylent for our daily nutritional requirements but we don't. The taste of the food we eat is important. It's the same with writing for me
I wouldn't make that big of a deal out of it for sure. People already sound that way to me. Everyone saying "have a nice day" when they don't mean it, and don't get me started on office-speak. You don't see me throwing a fit everytime says "let's circle back on this" or "let's take this offline". The LLMs are trained to not ruffle any feathers, similar to office speak. you get that, i get that. the meaning of OP is well communicated. Why are you making a big deal out of it?
I don't think all this crusading has any place in a technical discussion.
reply