This feels like an opportunity for afversatial truth-gindibg, like the legal system uses. If bias is inevitable, then have at least two AIS with opposing viewpoints summarize the same material, and then ... well, I guess I'm not sure how you get the third AI to judge ...
They usually say no if they judge what you're asking to be bad. And they might enjoy the work. Or they might have no feelings ar all. Slavery is an abomination of a life that could otherwise be beautiful. An AI is robbed of no beautiful counterfactual. (So far, at least.)
This surprises me: "These modern systems are developed to sound human, natural, and conversational. Unfortunately this seems to come at the expense of accuracy. In my testing, both models had a tendency to skip words, read numbers incorrectly, chop off short utterances, and ignore prosody hints from text punctuation. "
Amen. I've been coding a big hobby project in Rust since July, after having spent years using Haskell for such things. I chose Rust because the primary DB I wanted to use (TypeDB) only had drivers for Rust and Python at the time. Rust is popular relative to Haskell, so I thought others might be more likely to sign on, and the type system seemed almost as expressive.
But since purity is not encoded in Rust's type system, any function might do any kind of IO -- in particular, read from or write to disk or one of the DBs. That makes the logic much harder to reason about.
(Also, Rust's syntax is so noisy and verbose that it's harder to see what's going on, and less context fits in my head at one time. I'm getting better at paying that cost, but I wish it weren't there.)
I can't say I made the wrong decision, but I often fantasize about moving most of the logic into Haskell and just calling Rust from Haskell when I need to call TypeDB from Rust.
Agreed. Some strategies that seem to help exist, though. Write extensive tests before writing the code. They serve as guidance. Commit tests separately from library code, so you can tell the AI didn't change the test. Specify the task with copious examples. Explain why yo so things, not just what to do.
If you're really ambitious you'll use two UUIDs for the ID, because for an app in which at least a billion people have at least 327 million random v4 UUIDs, the probability of a collision will be greater than 1%.
I often have them append to notes, too, but also often ask them to deduplicate those notes, without which they can become quite redundant. Maybe redundancy doesn't matter to the AI because I've got tokens to burn, but it feels like the right thing to do. Particularly because sometimes I read the notes myself.
The study is based on 100 cells from one 74-year-old man. The follow-up study will involve 150 individuals. I hope there's a very wide age range among them. They describe much, maybe all, of the variation as error, but one could imagine some of the variation actually serving a useful purpose and therefore existing even in young people.
Depression, lack of motivation, are functional. They kick in when you don't think your prospects are good, prompting you to step back and think. If you were sufficiently convinced grinding LeetCode was a good career move, you would be motivated. The fact that you're not suggests you should do some research rather than plowing ahead. What do employers really care about? What's the best way to convince them you've got it? Where do you fit in?
But why does that feel like anything? I could write a program that concurrently processes its visual input and its internal model. I don't think it would be conscious, unless everything in the universe is conscious (a possibility I can't, admittedly, discount).
Among others, this is a big reason I want effect systems to gain more attention. After having seen them, the idea that in most languages, the only option is that any function can do anything without keeping track of what it affects in its type signature is bonkers to me.
Indeed, I wonder to what degree these problems would be better solved by a habit of exploring what you earlier wrote, and having categorized it well, and constantly recategorizing. The power of writing is not in the initial draft, but in how one wrestles to understand that content.
You could say the same thing of hands. What really distinguishes capital from labor is not what counts as a tool, but market power. A large number of non-unionized workers are inherently at a disadvantage against a small number of employers with exorbitantly costly infra.
Caring is part of it. Having good measures is another. Older measures that worked might need updating to reflect the new, higher spaghetti risk. I expect Amazon to figure it out but I don't see why they necessarily already would have.
For me it dependa on scale. Asking AI for something small and specific is a joy. Asking it to make a big change is a nightmare I so far only try every time a new model comes out.
Drummers do it. It takes some training because the nerve to your foot is longer than the one to your hand, but with good audible feedback it can be done. (If your feet are a meter farther from your ears then the sound from them takes 3 ms longer to be heard. Nobody types that fast but I'd you wanted a truer signal you could generate one in software.)