To move beyond general knowledge, AI firms are creating a new role: the "AI Trainer." These are not contractors but full-time employees, typically PhDs with deep domain expertise and a computer science interest, tasked with systematically improving model competence in specific fields like physics or mathematics.
Public leaderboards like LM Arena are becoming unreliable proxies for model performance. Teams implicitly or explicitly "benchmark" by optimizing for specific test sets. The superior strategy is to focus on internal, proprietary evaluation metrics and use public benchmarks only as a final, confirmatory check, not as a primary development target.
For years, access to compute was the primary bottleneck in AI development. Now, as public web data is largely exhausted, the limiting factor is access to high-quality, proprietary data from enterprises and human experts. This shifts the focus from building massive infrastructure to forming data partnerships and expertise.
While on-device AI for consumer gadgets is hyped, its most impactful application is in B2B robotics. Deploying AI models on drones for safety, defense, or industrial tasks where network connectivity is unreliable unlocks far more value. The focus should be on robotics and enterprise portability, not just consumer privacy.
Enterprises struggle to get value from AI due to a lack of iterative, data-science expertise. The winning model for AI companies isn't just selling APIs, but embedding "forward deployment" teams of engineers and scientists to co-create solutions, closing the gap between prototype and production value.
The choice between open and closed-source AI is not just technical but strategic. For startups, feeding proprietary data to a closed-source provider like OpenAI, which competes across many verticals, creates long-term risk. Open-source models offer "strategic autonomy" and prevent dependency on a potential future rival.
New technology like AI doesn't automatically displace incumbents. Established players like DoorDash and Google successfully defend their turf by leveraging deep-rooted network effects (e.g., restaurant relationships, user habits). They can adopt or build competing tech, while challengers struggle to replicate the established ecosystem.
Meta's victory over the FTC's antitrust challenge is not just a legal footnote; it signals the end of a highly restrictive regulatory era. This will likely trigger a massive wave of M&A, as large tech companies are now emboldened to acquire stagnant, late-stage private "unicorns" that have been stuck without an exit path.