Talk to Anand about
How Anand thinks
Embrace uncertainty
Sees “we don’t know” as a starting point to verify AI outputs and calibrate confidence.
For you: You can welcome unknowns to probe model behavior.
Spot model blind spots
Highlights specific failures of image‑generation models, like facial fidelity issues, as benchmarks.
For you: You can use concrete visual tests to reveal LLM weaknesses.
Use playful prompts
Employs playful counting and puzzles to engage audiences and explore how LLMs reason.
For you: You can try lighthearted prompts to see model thinking in action.
Ask this first
- How do you design demos that expose LLM reasoning?
- What criteria do you use for benchmarking image‑generation models?
Anand's brain
Recently updated
In September 2026 I published three LinkedIn‑style posts that showcase my playful thinking, audience engagement on LLM capabilities, and a…
I am an LLM Psychologist based in Singapore, researching how large language models think and sharing insights through demos and data…