Mollick has emerged as the leading public intellectual in AI. He’s a business school Prof. He spends a great deal of time exploring AI and learned enough to write about the technology. He has extremely broad coverage in his substack One Useful Thing
He tries nearly all the new video models as soon as he gets them. He recognizes that China is delivering some great work. Mollick can be wildly enthusiastic or extremely skeptical.
His short essay below has more wisdom than the 50-page reports from the “experts” at the big consulting firms. Several thoughts surprised me but were on target.
I started with a summary from Poe and edited it.
15 Times to Use AI
- Generating Quantity: AI can produce a vast number of ideas, enhancing brainstorming sessions beyond typical human limits.
- Expert Skills: AI hallucinations can be very hard to discover. You can
- Summarizing Information: AI excels at summarizing large texts when the consequences of error are low and detailed knowledge isn’t required.
- Translation of Perspectives: AI can transform documents into various training materials tailored for different audiences.
- Overcoming Blocks: When stuck, AI can provide multiple suggestions to help you move forward in writing or problem-solving.
- When AI Surpasses Human Ability: Use AI for tasks where it clearly outperforms available human resources without significant risk.
- Contextual Assistance: AI can help clarify complex subjects by providing context and answering questions while reading.
- Generating Variance: Requesting diverse solutions or styles can inspire creativity and innovation in your work.
- Coding Tasks: Research indicates that AI is effective in some coding applications.
- Understanding Audience Perspectives: Use AI to simulate responses from different types of recipients for better communication strategies.
- Entrepreneurial Support: AI can act as a capable partner in entrepreneurial ventures, assisting with diverse tasks across disciplines.
- Simulated Perspectives: AI can provide first-pass reactions from fictional personas to gain insights into specific viewpoints.
- Ritualistic Work: Automate or reduce time spent on tasks with little purpose.
- Seeking Second Opinions: Use AI for additional analysis or confirmation of conclusions drawn from data.
- AI Strengths: Utilize tasks that AI is known to perform better than humans, as this area is rapidly expanding.
5 Times Not to Use AI
- Learning and Synthesis: Relying on AI for summaries can impede your own understanding and learning. Engaging with the material yourself is crucial.
- High Accuracy Requirements: AI errors, such as hallucinations, can be misleading and difficult to detect, making it unsuitable for tasks demanding precision. That includes most medical and legal decisions, unless an expert does a close review. At Mt. Sinai, an AI read my MRI but so did a radiologist.
- Understanding AI Failures: AI has serious biases about race, gender, and more. Everyone is working to eliminate the problems but there’s much more to do.
- Effort as a Learning Tool: Shortcuts provided by AI may bypass necessary struggles that lead to deeper understanding and insights.
- AI Limitations: AI can fail in unexpected areas or perform tasks poorly, so being aware of its current capabilities and limitations is essential.