a slightly different metaphor. Copilot suggests it is next to you, helping you pilot... something else. The computer? The system? But "piloting the LLM" changes the relationship. The LLM is the thing that is being piloted.
The article is confused. The opinion is, it's so much safer _now_ than it was in the 1970s, it makes no sense to restrict children's wanderings.
But the article doesn't consider whether restricting children's wanderings is the REASON it is so much safer for children now.
"We have so many fire-safety rules in the building codes in Seattle. But get this: we haven't had any major fires since 1889! It's obvious we don't need these rules!"
It's true there is a cost to restricting children. But let's be a bit more realistic about the tradeoffs.
It's too simplistic to imagine the tension is between robot patrol dogs vs automating drudgery. If automating drudgery suddenly puts 30% of people out of work, it has huge broad negative impact on people who are currently alive and working in the current system. Innovate, but do it with awareness.
Your first counterpoint seems unnecessarily picky.
> So while it is a suitable DSL for many things (it is also seeing new life in web components definition), we are mostly only talking about XML-lookalike language, and not XML proper. If you go XML proper, you need to throw "cheap" out the window.
But the TWE did not embrace all that stuff. It’s not required for its purpose. And to call it “xml lookalike” on that basis seems odd. It’s objectively XML. It doesn’t use every xml feature, but it’s still XML.
It’s as if you’re saying, a school bus isn’t a bus, it’s just a bus-lookalike. Buses can have cup holders and school buses lack cup holders. Therefore a school bus is not really a bus.
I don’t understand how “running it in a vm” Or a docker image, prevents the majority of problems. It’s an agent interacting with your bank, your calendar, your email, your home security system, and every subscription you have - DoorDash, Spotify, Netflix, etc. maybe your BTC wallet.
What protection is offered by running it in a docker container? Ok, It won’t overwrite local files. Is that the major concern?
Yes and if you have o(1000) prs or issues in a single day, I guess you have a different scale of problem and will need to automate triage and management, probably through an ai-powered tool.
Even if it is 0.01% per day it’s still a very high volume. At that point it’s not a hobby. Millions of users maybe would point toward a more formal management and governance approach.
See Gemini-cli for example. Or a bunch of Microsoft projects. They use ai to triage and respond to tickets. (And they pocket veto many of them)
You can invest a lot and get minimal results, OR, it’s possible they invested in 71-odd tools and only 17% produced results, but those results were as desired or expected so they didn’t actually need the other 83% of the tools they tried.
The number of tools that were deemed effective is not proportional to the “the effect”.
You’ve asserted “THIS is not a solved problem,” which suggests everyone is clear on what THIS means. I think that is not a good assumption.