Well, you still need to decide if structured output is the right choice.

As they point out - this might impact results where deep reasoning is required.

So you might be better off taking the unconstrained approach with feedback.

The only "solution" with the unconstrained approach is to ask the LLM to regenerate the JSON. This is definitely more expensive than whatever downside from requesting structured outputs from the API.

ESPECIALLY with situations where deep reasoning is required, since those are likely to correlate with longer JSON outputs and therefore more failure points.

Definitely. A lot of what is missing in many discussions is the absolutely essential need to have evals.

The only way to “know” what is the best (or better) approach is to have a significant number of test cases that you can measure performance against.

At the moment, for a lot of people, state of the art is “let’s try a different prompt and see if the answer on my one example is better”

Here you go:

https://blog.atomic14.com

I would say “could not possibly be implemented” rather than “bold”.

Anyone can propose a brave or bold course of action. It’s very rare these people have any idea how to actually execute their plans.

> It’s very rare these people have any idea how to actually execute their plans.

Regarding Cummins, Why exactly? Dominic Cummins is articulate, seems to be quite intelligent and seems to be very fact/data orientated. I've also heard him describe how he would action particular policy.

Therefore I find it hard to believe he had didn't have any idea on how to execute his plans.

You are Dominic Cummins and I claim my 5 pounds :)

It seems when some people don't have an answer they prefer to deflect with a joke.

I think one issue we are having is that more and more things are said to be impossible to implement to the point that nothing happens... There is a lack of ambition, boldness, and leadership.

I don’t know.

Increasingly I see people offering simplistic solutions that don’t even pass basic smell tests.

And then when you point out the obvious flaws the response is that you just have to be brave or take a risk.

But I do agree - we seem to be in a world full of intractable problems and doing something may be better than nothing.

Yes there are simplistic solutions but, on the other hand, more often that not I think that claiming that issues are extremely complex is a way of avoiding doing anything for whatever reasons. So, it depends.

I think that the UK won't solve its issues until it gets a PM with a bold plan and great leadership, whatever side they may come from.

I would encourage you to do some more research. Maybe get out of your bubble. The world might not be as black and white as you think it is.

Definitely feeling my age. Growing up we had a "dripping bowl" that was kept in the fridge and received any left over fat from roasting meat.

This would then be used for frying etc.. I imagine my parents would have used it when they were young for "dripping" sandwiches.

Maybe this was just a UK thing?

It’s always important to remember what your position is when making off hand remarks.

An off the cuff comment to a friend or a colleague where you are both equal in stature/responsibility - probably fairly harmless. But important to also remember that you often don’t know what someone else is going through.

An off the cuff comment when you are the CEO or CTO to someone junior - potentially catastrophic for them.

The P4 is pretty high spec with a 400MHz dual-core RISC-V

Especially as there is a decent working BasiliskII port for the PlayStation Portable with its 333MHz single-core MIPS CPU.

So this should be much easier.

I wrote a clone of battle zone the old Atari tank game. For the enemy tank “AI” I just used a simple state machine with some basic heuristics.

This gave a great impression of an intelligent adversary with very minimal code and low CPU overhead.

Game design is filled with simple ideas that interact in fun ways. Every time I have tried to come up with complex AIs I ended up scrapping them in favor of "stupid" solutions that turned out to be more enjoyable and easier to tune.

I can vouch from my experience of turn-based games that exploiting a dumb AI often makes the game more fun (and gives the developer license to throw more/tougher enemies at the player), and noticing the faults really doesn't degrade the experience like you'd expect.

Unless enemies have entirely non-functional pathing. Then it's just funny.

I would reframe this as unable to rather than refused to.

It’s incredibly difficult in a successful business to change direction or even innovate.

There’s a bunch of guys in the engine room throwing coal into the furnace.

There might be one or two lookouts shouting - but they are drowned out by the noise of the engines.

If you’re lucky there’s a guy in the wheelhouse desperately spinning the wheel - but the connection between the wheel and the rudder is broken.

I think, for a lot of people, solving the problem was always the fun part.

There is immense pleasure in a nice piece of code - something that is elegant, clever and simple at the same time.

Grinding out code to get something finished - less fun…

It depends. Sometimes they joy is in discovering what problem you are solving, by exploring the space of possibilities on features and workflows on a domain.

For that, having elegant and simple software is not needed; getting features fast to try out how they work is the basis of the pleasure, so having to write every detail by hand reduces the fun.

Sounds like someone who enjoys listening to music but not composing or performing music.

Or maybe someone DJing instead of creating music from scratch.

Or someone who enjoys playing music but not building their own instrument from scratch.

No.

Building the instrument would be electrical engineering.

Playing the instrument would be writing software.