What happened to the artificial-intelligence revolution?

There is still impressive progress in large language models themselves every week. (Just to mention some of the past month: Claude-3.5-Sonnet, Gemma2, Nemetron and many more)

What is, in general, strange is that noone has really figured out how to do the plumbing. We have the LLM and they can perform almost any text related task, summarize search results or provide complex code snippets based on limited specification.

But somehow, the integration into work flows remains cumbersome.

- Search somehow seems to be burdened by the inability of the search providers to process entire webpages. Google, despite their search advantage, seems to only be able to process the search snippets instead of summarizing the entire content of websites. Most likely a copyright issue...

- Github Copilot is still basically autocomplete or a chat interface where I manually have to copy&paste results. Prompting for changes across multiple files is not really solved. (I know there is cursor, but my experience was quite mixed).

- All the hailed agents seem to create a lot of fluff but little actual code beyond what I would get with zero shot prompting. (Just tried a new tool today, which consumed $2.00 in API credits on the first task and left me with a broken codebase).

- Nothing that properly addresses slide generation yet?

Anthropics new workflow with Artifacts and Projects seems very promising and is a great leap forward. But it cannot natively process diffs or work with multiple soruce file and is therefore limited in total codelength.

As other people in this thread already remarked, maybe this is early stage technology that is pushed to commercialization too soon.

Amara's law : We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.

The AI revolution has only just started. The web took years to really find its feet.

Anyone who is calling AI a fizzer is calling the end of the race when the horses have just started out of the gate.

A simple skeptical summary of what the current generation of AI has accomplished to date—economically—relative to the huge investment in hardware and brains.

You have probably already read this kind of “money only” analysis (there was a better thinktank HN link last week on the topic of the massive investment overhang of about 500 billion dollars).

The Economist piece also asks (with almost tacit disappointment) why AI has not already allowed us to lay off many more employees.

This is only puzzling if the time-horizon of an analysis is measured in months (or at most a few years) and if you think of the world economy as a nimble gazelle.

There is plenty of work to keep humans busy and even satisfied with their lives. We just need a lot more human intelligence and cooperation to adapt to AIs impending impact. And keep it safe. It will make the 500 billion dollar investment overhang look like a small blip.

Over the last year, we have seen many publications in The New Yorker and The Guardian, even books, titled “Why artificial intelligence is not Intelligent or not Artificial?” that discuss the complexity of calling AI “Artificial Intelligence.”

The human mind still outdoes Chat-gpt, but there are much more advanced systems. What lies beneath the term “intelligence”? Does fearing AI only lead us to use it incorrectly? How can we use AI for our advancement?

Artificial Intelligence (AI) has become a buzzword in today’s technological discourse. The term "AI" has captivated the public and scientific communities alike, but it is essential to critically examine what AI represents.

The problem with the current situation is that AI was oversold. Everyone expected to see Cortana and all they got was SpongeBob. If the product wasn’t overhyped then perhaps people would pay more attention on how to utilize it in their everyday workflow without expecting spectacular results. Even a 5% boost in productivity is good long term.

> What happened to the artificial-intelligence revolution? So far the technology has had almost no economic impact

you could say this exact same thing about the internet for many years (e.g. mid eighties to early nineties) after it's first use.

I guess the influence on the economy may be one of these exponentially increasing things like 0.1% on GDP the 0.2%, 0.4%... so you don't notice much for a while but it gradually picks up.

I've noticed a kind of schadenfreude feeling coming from the mainstream media recently with these posts about what's going on with the profits of AI.

They, for sure, won't be getting any. So going against it is probably where they can squeeze the maximum of eyeballs they will get from all the "revolution".

You have all become AI food. Resign to your fate, past overlords.