API calls are expensive and slow, so I’ve shied away from integrating them into my personal projects. And of course there’s always the risk of hallucination.
GPT Chat seems smarter for business questions or planning that does not require iteration or wrangling data sources.
But Work gives the model a computer instance it can use for programmatic data access, API munging, etc.
Success cases I’ve had:
chat - read the reports in my Gmail and put a counter proposal together.
Work - cross reference all the options with commute time using traffic aware map data. Ground it on drop off / pick up times from each option’s stated rules and hours. Create a table with commute and idle time break downs.
That makes no sense. Do you think their plan is to spend billions of dollars and build infrastructure mega projects just to let it sit idle?
Did the rise personal computing make data centers and super computers obsolete?
Any advance in inference that allows local models to do the job will also benefit hyperscalers. Imagine the sheer amount of compute they could throw at problems if each 32GB of VRAM was enough for frontier reasoning.
They could also be betting that there will only be 1 or 2 big winners. And if they don't win they are saving on expenses, if they do win then they will have all the money they need to upgrade later if they want.
I totally agree about the subscription being worth it for most heavy users, but there are a lot of people in the developing world for whom $20/month is hard to stomach. I think it's good that those people have an option to get more access to AI with lower out of pocket price.
"couch all their agreements with caveats and provisos."
When you're a ChatGPT Projects or Claude Projects user, those caveats and provisos are your worst enemy because they'll change caveats into hard rules (either for the session or committed to memories) and you end up in absolute hell having to make it investigate to figure out why it can no longer produce anything but read-only pre-check code that never actually does anything but keeps performing stupid safety checks.
Yep the only way out is hooks to forbid what can be detected by ast and second model to prune comments, flatten pyramids of fallback, and squash the test suite removing quirks maintaining wanted behaviors.
This is commonly why, on Reddit in particular, you can get eaten alive.
Someone confident but incorrect, can often sound more convincing than someone with actual expertise. The expert must add caveats/hedge, because those are the facts on the ground, whereas the person reciting google can be entirely confident.
Of course the people judging aren't experts, so they side with confidence and simplicity. Heck, just writing shorter replies on Reddit is rewarded. Nobody reads the articles, let alone a paragraph-long reply.
That all being said though, there are limits. Sometimes LLMs on high-thinking go off on full tangents based on little, and don't have the self-awareness to bring it back.
I consider, on the contrary, caveating and hedging annoying 'typical redditor'/internet behaviors: they care more about being "technically correct" than conveying the message. On the internet, if you make even the tiniest mistake or simplification, someone will criticize you, so you're trained to always hedge. In normal discussions with friends you can just make general statements and people get what you mean.
You aren't arguing a "contrary" to what I actually wrote above. You're arguing with a strawman Redditor, and a point adjacent to what I was posting about - experts responding to topics within their actual realm of expertise.
So I won't be addressing this, for those reasons and others.
My experience has been very much the opposite of yours.
To an expert communicating with a layperson is a form of compression. You must turn some very complex idea into one that you suppose the other person can grasp given their limited frame of reference. It's always lossy, and you have to guess how much you can remove without sounding patronizing or being inaccurate. It's tough, and the more you know the tougher it gets.
Ever done that "explain what happens when I visit Google in my web browser" interview question?
A sales guy will answer in a sentence. An engineer might be able to talk about it for several days and still not be sure they didn't miss anything important. That much knowledge can actually be detrimental to communication.
What you're describing sounds precisely like the state of politics. The most successful politicians on both sides just scream "simple" solutions that don't stand up to even cursory examination. On the one hand, Trump, with 90% of his pronouncements being of that variety. And the DSA type, who just scream that we can fix everything if only we had much higher taxes (but ONLY for "billionaires" who would definitely pay them, perpetually, and wouldn't just leave).
Hardware is hardware. At some point you have to run and test it on the real thing, and it always misbehaves in some previously undiscovered way.
A lot of software people live under the illusion that hardware bugs don't happen, but they're ubiquitous and it's the job of firmware and OS people to reduce their impact.
Edit: I should add anyone that's worked with me knows I strongly advocate offline fuzz testing of all software components, especially to deal with unexpected states coming from the hardware side. This has proven unreasonably effective, and absolutely is something agents can help with, but they won't solve the whole problem.
It's been pretty ad hoc, but my prompts are nothing special. Things I generally do:
1. Top end model on high/xhigh thinking (last time I did it it was Sol xhigh I think)
2. Make sure it creates some representative fixtures of different sizes and sets up a good testing, profiling and benchmarking loop that doesn't require my input.
3. Make sure it has access to reference implementation code
Edit: Oh and one obvious pitfall that for some reason I still have to remind even smart models of from time to time: make sure it knows not to try to parallelize its benchmark runs. I've occasionally had an agent struggle to figure out absolutely nonsensical data because it tried to run multiple tests on the same compute hardware simultaneously.
Codex has full computer use capability. Surely such bans mean nothing even in the very near term.
reply