The year is 2030, my boss asks ChatGPT to do my job.
Result:
function doJob() {
// put the code to do your job here
}
He tweaks the prompt 42 times hoping to get a good result. He looks up the latest in LLMs, Oogabooga, suplex prompting, quantum recall, none of it works.
I really don't believe this generation of AI (transformers) is going to get us to what the article describes. They're great at certain tasks but are frustratingly dumb most of the time.
I've have worked with ML/AI tech for 15 years, implementing software.
I've worked very deeply with AI/LLM tech for past 3 years, implementing software.
I used GenAI/LLM tech every day at work.
And I can tell you absolutely you are correct. The limitations of the GenAI tech are way bigger factor than most realize.
Result:
function doJob() {
// put the code to do your job here
}
He tweaks the prompt 42 times hoping to get a good result. He looks up the latest in LLMs, Oogabooga, suplex prompting, quantum recall, none of it works.
I really don't believe this generation of AI (transformers) is going to get us to what the article describes. They're great at certain tasks but are frustratingly dumb most of the time.