What about opensource self hosted AI?
There are still major ethical issues with how they get training data, but the problem with hijacking the economy and society seems resolved. I feel like I can more easily forgive the copyright theft for someone just fucking around in their homelab, replacing Google assistant.
Maybe I’m just bean-souping.
As much as it pains me to carry water for AI, do you think that people are replacing things they liked to do with AI?
Like… They were reading for pleasure, and then they had the AI read it for them instead?
That they were talking to their friends they enjoy speaking with, and then they had the AI talk to their friends instead?
Or are you saying that what you describe is the experience that you get from reading technical manuals for your work? For reading warning labels? From talking with coworkers you dislike?
I agree with you for common CLIs that the model will natural have in its training set, but when building your own CLI it seems like the self-documenting nature and progressive disclosure of MCP would have a huge advantage.
With a cli you don’t have the prefix cost, but your agent also doesn’t know that the tool exists, what it’s for, how to use it, etc. once you give it that information, then you have the same issues with prefix cost, right? Changing tools does break the prefix cache, but changing tools is a thing you can do. You can’t change (model awareness of) CLI, and if you could then it’d also break the prefix cache, or you’d have tool instructions scattered throughout the context without structure.
Dumping to file is a harness thing, not a MCP thing. A harness will do the same with large terminal outputs.
Likewise setting up a new CLI can be complicated. I never see an agent try to install an mcp, but I also don’t typically see it try to install a new CLI.
I don’t see how mcp is any more a fixed set of tools than CLI is. It’s just presented to the agent in a structured way, which is important for discovery and instruction following.
But yeah ease of chaining is certainly useful wrt CLI. That’s a huge win for sure.
Can you elaborate on LLMs using CLI more effectively than MCP? Do you have any sense of how likely it is just your domain or the MCPs you’re using? Do you have any theories about why that may be?
Also same thing but no straight men.
This kinda gets philosophy-of-cognative-science-ey, but it kinda smells like “measuring a fish by its tree climbing ability”. Or maybe like goodharts law.
There are many different kinds of smart, is the kind of smart they’re measuring the one that will be relevant?
Again, I don’t want to carry too much water for AI, but I also don’t want to propagate arguments that also apply to other tools I like. Consider: externalizing memory. Without the tool you used to capture that memory, you’re completely unable to function. Dependency on it reduces your ability to remember things yourself. By many measures, it makes your dumber and less productive. But I wouldn’t personally consider taking notes to be detrimental.
So I want to be very careful when we make claims that externalizing some aspect of our cognition makes us worse somehow.