The microwave oven is a strange choice of metaphor here, considering that even now microwaves sell 100's of millions of units every year, are nearly ubiquitous in households in the western world, are present in nearly every commercial restaurant and kitchen outside fine dining, and reached $5m/year in sales in their first decade, $30m/yr in their second, and doubling basically every decade since, before finally reaching near 100% market penetration and plateauing around $15b year ever since.
I mean, I get the author's point not to over-hype AI, but the microwave oven is one of the most successful inventions in the past 100 years.
How about ... the Segway? I hear whole cities will be designed around them.
Microwave oven may not be the best analogy but its a better analogy than Segway because both microwave ovens and LLMs are both definitely not failures. Millions of people pay to use LLMs now. Certain jobs have been entirely eliminated by LLMs (media back office notable - dividing video into distinct scenes, etc).
So the point is even successful new technology can be over-hyped.
Of course, LLMs may or may have a functionality that expands to everything also. Unlike a microwave over, if LLMs have limitations, those limitations aren't visible.
1.) customers who don't know about the practice, now expect food right away for very cheap price, which bankrupts traditional restaurants that cook from fresh ingredients. An analogy could be off the shelf software or custom software losing sales to vibe coded software
2.) customers who do know about the practice, stops ordering food altogether, due to food issues (prepared ingredients are very low quality, oil could be very dirty) or they could cook prepared food at home in their own microwave for even cheaper. An analogy is programmers coding their own software using vibe coding.
3.) declining quality of food, and less and less people eating out in China. It's gotten so bad that now hotel restaurants, which would be the fine dining options, are setting up food stalls in the street to sell cheap but freshly cooked food, in order to get people into the hotel restaurants. An analogy might be softwares with per seat pricing changing to action based pricing for an initial period.
The speed oven has kind of taken over a lot of food service. It's how Starbucks offers hot meals, for example. But there will still be a use for steam-jacketed kettles and fryers.
I think LLM is here to stay but cheap output won't have value. We're rapidly tired if its flooding of the zone with spam. The microwave killed some traditional ways of cooking, and the LLM is killing some older ways of communicating. Not because it's better, but because it makes e-mail tiresome.
Almost all the cooking I do are using microwave. While they are not bad, there are certain critical steps microwave cannot do. For example enrich the flavor of garlic and other spices in hot cooking oil. I found microwave is not good at heating vegetable oil and the garlic is more like boiling in lukewarm oil instead of frying. Also it heats food inside out equally as compared to stove which transmit heat from the out layer to inside, as result the food have different textures.
The restaurant last night cooked my chicken in the microwave. It would have been far better for everyone if before placing the plate in the microwave they scraped it into the trash. Then it wouldn’t have ruined my otherwise mediocre salad.
In the future where all food is cooked in the microwave I’m quitting. I’ll go construct a log cabin in the woods and cook my own food over a fire or go without.
17 comments
[ 349 ms ] story [ 1962 ms ] threadI mean, I get the author's point not to over-hype AI, but the microwave oven is one of the most successful inventions in the past 100 years.
How about ... the Segway? I hear whole cities will be designed around them.
So the point is even successful new technology can be over-hyped.
Of course, LLMs may or may have a functionality that expands to everything also. Unlike a microwave over, if LLMs have limitations, those limitations aren't visible.
1.) customers who don't know about the practice, now expect food right away for very cheap price, which bankrupts traditional restaurants that cook from fresh ingredients. An analogy could be off the shelf software or custom software losing sales to vibe coded software
2.) customers who do know about the practice, stops ordering food altogether, due to food issues (prepared ingredients are very low quality, oil could be very dirty) or they could cook prepared food at home in their own microwave for even cheaper. An analogy is programmers coding their own software using vibe coding.
3.) declining quality of food, and less and less people eating out in China. It's gotten so bad that now hotel restaurants, which would be the fine dining options, are setting up food stalls in the street to sell cheap but freshly cooked food, in order to get people into the hotel restaurants. An analogy might be softwares with per seat pricing changing to action based pricing for an initial period.
In india such clay pots are sold by villagers and are slowly getting in vogue.
With gen AI, it’s much less clear cut what the long-term limitations are.
I think LLM is here to stay but cheap output won't have value. We're rapidly tired if its flooding of the zone with spam. The microwave killed some traditional ways of cooking, and the LLM is killing some older ways of communicating. Not because it's better, but because it makes e-mail tiresome.
I love microwave, but the usual "if all I have it's a hammer..." is a thing
https://www.finalroundai.com/blog/github-ceo-thomas-dohmke-w...
HN comments thread, https://news.ycombinator.com/item?id=44797264