How I Use "AI" (nicholas.carlini.com)

Article summary

The author, a security researcher and programmer, shares their experience with using large language models (LLMs) to improve their productivity, citing examples of how LLMs have helped them with tasks such as writing code, automating monotonous tasks, and learning new technologies. The author argues that LLMs are not overhyped and have provided the single largest improvement to their productivity since the internet was created. The author also acknowledges the limitations and potential negative consequences of LLMs, but focuses on their usefulness in the present. The article provides a list of 50 conversations the author had with LLMs to demonstrate their utility.

Main themes

• LLM usefulness
• Productivity improvement
• AI skepticism
• Ethical concerns
• Technological limitations
• Code generation

What commenters say

• Some commenters agree that LLMs are useful for tasks such as writing code and automating monotonous tasks, but also acknowledge their limitations and potential negative consequences.
• Others argue that LLMs are overhyped and can produce suboptimal or even dangerous results if not used carefully, and that their outputs should be thoroughly verified.
• A few commenters express concerns about the environmental impact of training and using LLMs, and argue that their benefits may not outweigh their costs.
• Some argue that LLMs can be a valuable tool for learning and productivity, but only if used in conjunction with human expertise and critical thinking.
• There is disagreement about the extent to which LLMs can replace human programmers, with some arguing that they can automate many tasks, while others believe they are not yet ready for production-ready code.
• A few commenters note that LLMs can be useful for generating code, but that this code may require significant modification and verification before it is production-ready.
• Some argue that the benefits of LLMs are not limited to programming, but can also be applied to other areas such as writing and research.
• There is a concern that the increasing reliance on LLMs may lead to a loss of human skills and expertise, and that this could have negative consequences in the long term.