Why I'm Quitting AI

Summary

Michael Sturgill explains why he is stepping back from generative AI after using language models for years. AI initially felt exciting and useful for casual conversations and quick coding experiments. However, hallucinations, inconsistent results, and the gap between marketing promises and actual performance have made him less enthusiastic.

He compares AI’s reliability to virtual reality: even an occasional failure can make an experience frustrating. Similarly, generative AI can be difficult to trust when it does not reliably produce the result a user expects.

Experiences with AI

• Photography and AI retouching: After 15 years as a photographer, Sturgill tried AI tools in Luminar Neo. They sometimes missed blemishes or removed details he wanted to preserve, such as a person’s mole. He found manual editing more controllable and satisfying.
• Website copy: He used ChatGPT to draft copy for his website, but later found it generic and unlike his own voice. He plans to rewrite it himself.
• Sponsors and creative services: He is uncomfortable with Artlist’s move toward generative AI and is unsure whether its music and stock footage are human-made. He also says many sponsorship offers he receives come from AI companies.
• Game development and coding: Sturgill recalls spending days building an inventory system in Unity and values the satisfaction he felt when it finally worked. In another project, an AI-assisted coding attempt broke his Unity project, which he recovered using Git. He argues that working through coding problems can build skills and inspire creative solutions, while delegating the work to AI may take away some of the learning and fulfillment.

Perspective and plans

Sturgill’s central argument is that constraints and struggle can foster creativity and improvement. He would rather focus on learning and enjoying the process than simply producing an end result faster. He plans to cancel his AI subscription and write his own code, even if it takes longer.

He acknowledges that AI tools may become more worthwhile in the future, but says they currently take the enjoyment out of his work.

Other concerns

Sturgill also raises concerns about rising computer-part prices and the possibility of future consoles costing more than $1,000. He criticizes Sam Altman’s reported suggestion that people should not keep RAM in home PCs and that it should instead be used in data centers.

Main speaker and references

• Michael Sturgill: The video’s sole speaker, discussing his experiences with AI, photography, website writing, sponsorships, and game development.
• Referenced products and organizations: Luminar Neo, ChatGPT, Artlist, Unity, Git, and Sam Altman.