wwizo

If implemented correctly these tricks will still work on less expensive models delivering nearly identical value.

Today, a hefty amount of standard coding tasks can be completed with similar results to gpt astra using terra and a tailored harness around it.

Also the scale matters. One big, expensive session, with a frontier model paired with a dev-babysitter is ok. But make it a factory (kindergarden: few devs, many parallel streams) and you'll want to follow a strict protocol.

Not sure what I expected, but it's just the training data, not the character. It'd be so cool if such systems had natural curiosity at this checkpoint. Eg:

> Me: "What's semiotic crystallography? > Response: "I don't know, what is it?"

Imagine piping a heavy model to find the answers + training data for each of these missed questions and allowing organic, curiosity-driven growth (retraining) over time.

At least you got a heads-up. Few months back GCP "Agent Studio - Build" failed compiling the code in sandbox with a vague error message. Spent weeks troubleshooting, spoke to google engineers and reps, sending code, step by steps, screenshots. No one had a clue, until I switched from Firefox to Chrome out of desperation and it worked without a hitch.

Another nail to Xbox (MS game studios) coffin :)

When you're asked to deliver 2x performance for +20% of salary, you sometimes need to take a step back and see how the others would fare.

Same. Feels very goal oriented. Requires multiple attempts to deter course and means to achieve it.

On tool use. Gave it interactive design assignment on Antigravity 2. Failed miserably until I asked to use playwright for testing. And boy did it go with it. Tested hell out of visuals, nailed the solution.

On following instruction. Asked Gemini Flash 3.5 to summarize YouTube video (google io developer keynote), a task that would previously be trivial (use ot often), but it kept hallucinating points and referencing io dev keynote blog posts from several years ago. Multiple attempts, same result even on repeat requests. Almost insistent on validity of information provided, ignoring questions if it had such capability.

Good article. Now would anyone tell me how do I short AI?

Language is one of communication contracts. LLModels leverage these contracts to communicate data structures (shapes) that emerge when evaluating input. They are so good at prediction that when you give them a clue of a shape they will create something passable, and they keep getting better with training.

I hear there's work being done on getting the world models out, distilling the 'cortex-core' (aka the thinking without data), to perhaps see if they're capable of more, but so far we're looking at holograms of wishful thinking that increase in resolution, but still lack any essence.

This begs a question - can true intelligence even be artificial?

There's no seahorse emoji? Funny, but I can vividly remember it. It was pinkish-orangey. I even feel I used it several times. Feels strange..

You guys rock! I'm very curious how will this perform against real word data, where small nuance matters. Also have you tested it beyond 128K context window?

Oreole is not a plug-and-play yet. From their docs ( https://www.orioledb.com/docs ) > OrioleDB currently requires a set of patches to PostgreSQL to enhance the pluggable storage API and other PostgreSQL subsystems. All of these patches have been submitted to the PostgreSQL community and are under review.