AI isn't useless. But is it worth it?

🌈 Abstract

The article discusses the author's perspective on the usefulness and limitations of generative artificial intelligence (AI) tools, such as large language models (LLMs). The author acknowledges that these AI tools can be useful for certain tasks, but argues that the benefits do not justify the significant costs and harms associated with their development and deployment.

🙋 Q&A

[01] Use Cases of AI Tools

1. What are some of the use cases the author has found for AI tools?

• The author has found AI tools useful for:
  • Helping to find the right word when writing
  • Proofreading and catching typos or grammar mistakes
  • Generating simple code snippets or boilerplate for coding tasks
  • Summarizing meeting notes or other documents

2. What are the limitations of AI tools for these use cases?

• The author notes that while AI tools can be helpful for these tasks, they are not a replacement for human capabilities:
  • They can't replace an experienced developer for complex coding tasks
  • They sometimes introduce errors or non-functional code
  • Their writing still lacks the nuance and creativity of human-generated content

3. What other use cases are people finding for AI tools?

• The author cites examples of people using AI tools to:
  • Generate meeting summaries, email drafts, and other routine business documents
  • Create subtitles and transcripts for videos and talks
  • Assist non-native English speakers with professional communications

[02] Costs and Harms of AI

1. What are some of the significant costs and harms associated with AI development?

• The author highlights several major issues:
  • The high energy and computing costs of training and running these models
  • The exploitation of "ghost labor" - underpaid contractors doing data labeling and content moderation
  • The use of AI to generate disinformation, harass people, and enable increased workplace surveillance
  • The opportunity cost of using AI tools instead of hiring human workers

2. How does the author view the tradeoff between the benefits and costs of AI?

• The author believes the benefits of AI tools, while real, pale in comparison to the significant costs and harms. They argue that the tech industry's "innovation at any cost" mindset is unjustified.

3. What is the author's overall perspective on the value of AI tools?

• The author acknowledges finding AI tools useful in their own work, but ultimately believes the technology is being overhyped and that the costs outweigh the benefits. They call for more honesty about AI's limitations and a greater consideration of its negative impacts.