The currently hot LLM technology is very interesting and I believe it has legitimate use cases. If we develop them into tools that help assist work. (For example, I’m very intrigued by the stuff that’s happening in the accessibility field.)
I mostly have problem with the AI business. Ludicruous use cases (shoving AI into places where it has no business in). Sheer arrogance about the sociopolitics in general. Environmental impact. LLMs aren’t good enough for “real” work, but snake oil salesmen keep saying they can do that, and uncritical people keep falling for it.
And of course, the social impact was just not what we were ready for. “Move fast and break things” may be a good mantra for developing tech, but not for releasing stuff that has vast social impact.
I believe the AI business and the tech hype cycle is ultimately harming the field. Usually, AI technologies just got gradually developed and integrated to software where they served purpose. Now, it’s marred with controversy for decades to come.
If we develop them into tools that help assist work.
Spoilers: We will not
I believe the AI business and the tech hype cycle is ultimately harming the field.
I think this is just an American way of doing business. And it’s awful, but at the end of the day people will adopt technology if it makes them greater profit (or at least screws over the correct group of people).
But where the Americanized AI seems to suffer most is in their marketing fully eclipsing their R&D. People seem to have forgotten how DeepSeek spiked the football on OpenAI less than a year ago by making some marginal optimizations to their algorithm.
The field isn’t suffering from the hype cycle nearly so much as it suffers from malinvestment. Huge efforts to make the platform marketable. Huge efforts to shoehorn clumsy chat bots into every nook and cranny of the OS interface. Vanishingly little effort to optimize material consumption or effectively process data or to segregate AI content from the human data it needs to improve.
The currently hot LLM technology is very interesting and I believe it has legitimate use cases. If we develop them into tools that help assist work. (For example, I’m very intrigued by the stuff that’s happening in the accessibility field.)
I mostly have problem with the AI business. Ludicruous use cases (shoving AI into places where it has no business in). Sheer arrogance about the sociopolitics in general. Environmental impact. LLMs aren’t good enough for “real” work, but snake oil salesmen keep saying they can do that, and uncritical people keep falling for it.
And of course, the social impact was just not what we were ready for. “Move fast and break things” may be a good mantra for developing tech, but not for releasing stuff that has vast social impact.
I believe the AI business and the tech hype cycle is ultimately harming the field. Usually, AI technologies just got gradually developed and integrated to software where they served purpose. Now, it’s marred with controversy for decades to come.
Spoilers: We will not
I think this is just an American way of doing business. And it’s awful, but at the end of the day people will adopt technology if it makes them greater profit (or at least screws over the correct group of people).
But where the Americanized AI seems to suffer most is in their marketing fully eclipsing their R&D. People seem to have forgotten how DeepSeek spiked the football on OpenAI less than a year ago by making some marginal optimizations to their algorithm.
The field isn’t suffering from the hype cycle nearly so much as it suffers from malinvestment. Huge efforts to make the platform marketable. Huge efforts to shoehorn clumsy chat bots into every nook and cranny of the OS interface. Vanishingly little effort to optimize material consumption or effectively process data or to segregate AI content from the human data it needs to improve.