Tech projects failing is normal, what’s more telling is comparing them to non-AI tech projects.
But according to research by the RAND Corporation, over 80% of these AI projects will fail — which is twice the failure rate for non-AI technology-related startups.
Remember kids, the “failure” is economic failure, not moral, not utility.
Chatgpt is a “success”. If the company is stealing your face, art, voice but it is making money, they list it as a success.
By that metric, FTX was a “success” until their fraud was revealed.
80% fail economically, how many of those 20% fail morally? How many of those 20% have real utility? (E.g. not only generating weird picture of poor children building Mickey mouse out of bottles)
You don’t need to tabulate the moral failure. That’s 100%.
Billions in capital should be billions in taxes.
the internet can do really good things but for every Amazon we also had pets.com in the 00s burst i think we are speedruning the bubble with AI. I can’t wait till it pops so we can move on to something useful and stop wasting power.
I think we are in the beginning phases of ai. It has already changed how we do many things, but we are still looking how to use the tech, how to integrate it, how to make products. All while the tech is rapidly evolving.
This looks like a crypto bubble.
20% success rate, that’s pretty high to say it would ever go away. 80% of restaurants close in the first five years, that bubble hasn’t popped.
But it has historically over the past 50 years been dominated by franchised chains. Which is to say…terrible.
Agreed, which inevitably means chains would lower the rate of failure as well, so I imagine first timers fail at higher rates : / I always regret that I don’t eat at more hole in the walls.
and yet crypto recieves speculative investment to this day. Say what you will about AI, at least the end product can sometimes do something.
Most pieces of software fail (I’ve heard numbers in the range of 75%) according to the people who made them. 80% is only marginally higher than that, so … 😉
Literally addressed in the first paragraph. They say 80% it’s twice the normal failure rate
And yet other studies have the people making the software say it’s a failure 75% of the time. (Keep in mind that the 75% comes from the people making it, not the people selling it who have a vested interest in claiming their product isn’t a failure.)
Remember that AI is an extension of human understanding, knowledge and interpretation. Not a replaceable tool
AI has existed for less than a decade. We are fine without AI. You can’t just look at something still being built, with loads of problems, and call it irreplaceable. Nobody depends on AI for their survival, they depend on water, clean air, and other resources that AI is taking away. It’s also taking people’s jobs that, again, they need for their survival.
Stop emulating Elon Musk and get some sleep.
Uhhh, AI has been around for DECADES. You are likely conflating AI as a whole with “generative AI” which has been experiencing a boom the last 5+ years. Some examples of AI include pathfinding (think of your GPS telling you how to get somewhere), chess engines (also a form of pathfinding, incidentally), and NPCs in video games making decisions based on a set of rules and states.
Oh. Well then, replace “AI” with “generative AI”. I may have been a bit confused there. In that case, we are perfectly fine without generative AI, and it has guzzled resources and manpower. For such a cost, what is the return? You can look at this Hill article, or you can look at fake studies, or other ways GenAI has been harmful. Stuff like the A* search algorithm didn’t require jumbo subsidies for it to work properly, and it isn’t making the internet as it is worse.
So that’s what I’m saying. I have yet to find an argument proving, successfully, that generative AI is worth the cost.
Yeah, I definitely agree with the sentiment. I try to evangelize the fact that Gen AI !== AI because AI as a whole has a ton of amazing benefits at low cost. And it’s way more prevalent than people realize. But I don’t disagree that Gen AI has been appropriated for some of the dumbest shit ever that costs an arm and a leg to create. A* is a good example of what I was referring to in my original comment. Simple, effective, useful. That being said, I’m sure even Gen AI could have some worthwhile benefits if used in better ways, but it sure does seem like we can’t have many nice things as of late (2024) 🙃
