> “the specific advantages of AI must be obvious and worthwhile to justify the investment.”
I feel like this sums up a lot of what’s going on here. Tons of students immediately saw that ChatGPT could write essays, and bought in. As LLMs and agents have gotten smarter, they’ve become more useful for coding, and developers have started to get into them.
To give a counterexample: Facebook’s rollout of Meta AI was busted and aimless. Their search UI is bloated with suggested prompts for things no one wants to do like “Write a script for a pop culture podcast” or “Imagine dog yoga”. Many posts are followed by suggested prompts for weird questions about the post’s content, like “What are fun activities to do with cats?” and “What do cats like to eat?” after a picture of my friend with her cat.
I’ve started using LLMs a lot more the past couple months, and I can start to see the faint outline of what Meta was trying to do. But after burning their first impression on an underskilled LLM and undercooked product thinking, I don’t know that users will give Meta AI a second chance even if their features do get better.
> “the specific advantages of AI must be obvious and worthwhile to justify the investment.”
I feel like this sums up a lot of what’s going on here. Tons of students immediately saw that ChatGPT could write essays, and bought in. As LLMs and agents have gotten smarter, they’ve become more useful for coding, and developers have started to get into them.
To give a counterexample: Facebook’s rollout of Meta AI was busted and aimless. Their search UI is bloated with suggested prompts for things no one wants to do like “Write a script for a pop culture podcast” or “Imagine dog yoga”. Many posts are followed by suggested prompts for weird questions about the post’s content, like “What are fun activities to do with cats?” and “What do cats like to eat?” after a picture of my friend with her cat.
I’ve started using LLMs a lot more the past couple months, and I can start to see the faint outline of what Meta was trying to do. But after burning their first impression on an underskilled LLM and undercooked product thinking, I don’t know that users will give Meta AI a second chance even if their features do get better.
https://archive.is/g4lvY
Everywhere I interact with AI I am forced to interact with AI, and the service is worse than before AI.