models are the most expensive, the slowest, & the least capable that they'll ever be context windows are only going to get larger, models are just going to get cheaper, faster, more capable what could you build with infinite context, zero latency, no cost, unlimited tools use? – at Sunnyvale, CA

We've also seen strong adoption of genAI in Google, for every part of employees' work (answering emails, analyzing data, summarizing bugs, automating code reviews, drafting docs, etc). "Getting the job done" is going to look completely different–for everyone–over the next year.

🙌 Proud to be a part of this team, and to boost this number even more in 2024! We're just starting. – at Morro Bay, CA

as generative models are quickly becoming commodities, companies will need to differentiate on user experience, distribution, and building strong data flywheels (meaning: How well can you attract and retain new users? How many landing surfaces do you have to capture data? What is the quality of those data? How well can you translate those user insights back into pretraining/fine-tuning your foundational models? etc.) it makes me wonder if we'll see a resurgence in developer relations; hyper-specific, personalized marketing; and UX roles (and eventually, services using genAI to fully automate those activities) // shower thoughts – at Palo Alto, CA