perhaps. I know my prompts are like three paragraphs long lol. and then I do three our four iterations to clean things up.
When I messed with vibe coding over winter break, the basic approach was to create a really solid requirements document first. You go back and forth with the AI about what you want for a minimum viable product, appropriate data structures, the order of the code, and how to test it. And you check repeatedly if anything is missing or should be rearranged, before asking it to build anything. Itās basically the same approach that a human team should use. Then while the AI builds, it can reference the plan, and you can iterate.
Iāve read complaints about AI that it means young people wonāt learn to think and communicate in an orderly and logical manner.
It seems to me that writing out lengthy āpromptsā for AI requires thought and accurate communication.
Iām so old that I worked with actuaries who wrote āspecsā for programs that programmers would write. AI prompts for larger jobs sound similar.
1 Like
by the way, that was a claim dispute with Liberty Mutual.
It was Liberty Mutualās lawyers who were sloppy
1 Like
Happy new year everyone!!!
I havenāt been following humanoid robots, so this is new to me.
It is impressive.
Looks like we now have the ability to run four different AI models at the same time via the same prompt pathway.
This is a good idea. Will cut down on the weaknesses of each of the models in specific areas.
Perplexity - Claude - ChatGPT - Gemini
So we use AI to run AI to get better results?
1 Like
I just saw this:
Distressing.
Iāve been treating AI like my assistant at work rather than trying to outsource any key parts of my job. Itās been going pretty well.
Before a meeting, Iāll have it come up with topics I should prepare for or questions I should ask. Iāve provided it with pitch decks and asked it to break it down for me to the key components. This has allowed me to focus the conversation on what matters most.
I have asked Copilot to remember some key metrics I am evaluated against and will have it connect those things to my meetings or projects. Iāve used it to go through my email to help me write my year end self evaluation. Iāve had it analyze my calendar to identify areas I can be more efficient.
Is it groundbreaking? Not at all. Is it making me more focused in my role? Yes. Iām not asking it to make a strategic decision, Iām still owning that. But I have more time to do that stuff now.
2 Likes
Can it write emails and Teams messages responding to my bossā idiotic questions and requests?
Automatically, I mean. So I never have to actually see or read them.
4 Likes
Something tells me that you ought to read/edit anything from AI before sending it to someone with power over your paycheck . . .
You could include in the prompt - āWrite it in the style of a straight shooter with upper management written all over themā
11 Likes
The more I hear AI execs speak, the more I think they just dislike people in general
I struggle to understand why people listen so much to Sam Altman. Heās not particularly honest and forthcoming, heās not super-intelligent, and heās not a cult of personality like Musk. Heās just the guy who happens to be in charge. Kind of like Zuckerberg, except nobody cares about Zuckerberg?
Sometimes i think we have to suppose the uber rich are also uber smart because it justifies our belief that we live in a meritocracy.
I think it peeves me in particular because the AI world is loaded with actual geniuses. Some of them are uncharismatic, but Sam Altman also isnāt charismatic.
Open AI being called in for a chat with the AI Minister.
https://www.cbc.ca/news/politics/open-ai-summoned-ottawa-tumbler-ridge-9.7103281
I get theyāre in an awkward spot here where they have to balance between privacy and protecting the public. Iām not really sure where the line is on this.
2 Likes