PINAKI LASKAR

I do not know what I may appear to the world, but to myself I seem to have been only like a boy playing on the seashore, diverting myself in now and then finding a smoother pebble or a prettier shell than ordinary, whilst the great ocean of truth lay all undiscovered before me.

A tech company led by people who don't understand the technology is not a tech company. It's a sales organization.

AI for good is not inevitable. AI for good can only be stewarded by the way that we structure them in systems, the way that developers use them, the way that we build guardrails and laws around them.

The generation that refused to accept cookies, is now giving Al access to their desktops, and bank accounts. 𝗧𝗵𝗲 𝘀𝗮𝗺𝗲 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝗼𝗻 𝘁𝗵𝗮𝘁 𝗳𝗼𝘂𝗴𝗵𝘁 𝗮 𝗰𝗼𝗼𝗸𝗶𝗲 𝗯𝗮𝗻𝗻𝗲𝗿 𝗷𝘂𝘀𝘁 𝗴𝗮𝘃𝗲 𝗮𝗻 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁 𝘁𝗵𝗲 𝗸𝗲𝘆𝘀 𝘁𝗼 𝘁𝗵𝗲𝗶𝗿 𝗳𝗶𝗹𝗲𝘀.

Understanding AI Atrophy, When The Tool Uses Us. AI Atrophy does not mean AI makes people “stupid.” It means repeated delegation of cognitive effort can weaken skills that are no longer exercised. Used well, AI is a scaffold. Used passively, it becomes a crutch.

Your AI is lying to your face, And it's doing it on purpose. linkedin.com/pulse/ai-sycop… #AISycophancy

How companies interpret "human in the loop".

Our scientific power has outrun our spiritual power.

🧠 When you rely on AI too much: → You stop struggling (where real insight comes from) → You accept first answers without questioning → You build faster, think shallower. ⚠️ AI gives answers. Not judgment. 💡⚡

Anthropic just released a fascinating study on WHICH JOBS AI IS ACTUALLY REPLACING and which ones it’s merely assisting.

Can you think of an answer? Name a word that can be read the same upside down if it’s typed in all capital letters…

Because vibe coding has matured to a point that it is impossible to ignore. And it can be experienced by everybody with very modest effort. Any executive who does not at least spend one day exploring what is possible this year is failing their organisation.

ℹ️

AI systems don’t own Reality. Most AI failure narratives focus on: - Epistemic failures: The model didn’t know enough; - Ethical failures: People didn’t care enough; - Compliance failures: Not enough policies or oversight; But the real failures are architectural.

For those deploying LLMs in regulated industries : Legal, Finance, Healthcare - Fluent text is not enough. You need measurable reasoning depth. Long Answers Don’t Mean Deep Reasoning, How to Measure LLM Reasoning Depth in High-Stakes Domains.

AI Governance is Not a Safety Problem. It is a Sovereignty Problem. At that point, humans are supervising not governing. Supervision is not sovereignty. If governance does not exist architecturally at runtime, if human authority can be optimised away, then control has shifted.

Compliance is not a PDF, It is infrastructure. We are building a governance API designed to move you from black box uncertainty to verifiable operational control. Apply to the Act or shut down your AI.