We train doctors for years, then trust them with our lives. We raise kids, then let them make their own decisions. AI agents are next — and they'll need the same thing. Coaching. At World Computer Day in Davos, Prof. Ramesh Raskar from MIT made a prediction most people aren't ready for: The next big job title won't be AI Engineer. It'll be AI Agent Coach.
Cloud lock-in is the tech industry’s dirty secret. Cloud engines on @dfinity are about to fix that: → Create an engine on Amazon → Deploy your apps → Move the whole thing to Google — instantly → Or distribute across both for resilience Just like the internet runs across everything, the Internet Computer will too.
A moment of appreciation for Arjaan Buijk (@icpp_pro), one of the ICP ecosystem’s earliest and most dedicated builders, who: - discovered and got hooked on ICP 1.6 months after its genesis - got the first LLM working in a canister - went on to co-found @onicaiHQ, together with Patrick Friedrich (@thepatnorris) They were the first to formalize and popularize the concept of “Proof of AI Work” for decentralized AI networks. 1/2
The brain is still the only AGI system we know of. It can generalize beyond its training data. Adapt in real time. Learn from just a few examples. Today's AI? Mostly brute force engineering and curve-fitting. Prof. Andreas Tolias (Stanford) on what keeps him up at night — finding the "Newtonian mechanics of intelligence." His aim? Extract the brain's principles and bring them back to AI.
“2026 is the year the Internet Computer truly connects with the mass market.” After years of development, ICP is pivoting from developing infrastructure to productization. Caffeine proved the appetite exists. Cloud engines are the next step. Massive scaling incoming.