Jack Clark

Elon Musk

2026-10-01 04:35:21

Musk, Jack Clark and Jaime Teevan point students toward broad education in the AI era

Three tech figures from different corners of the industry are giving young people a similar answer on what to study in the age of artificial intelligence: do not overcommit to a single specialty too early, and build a broad base of knowledge instead. Tesla CEO Elon Musk, Anthropic co-founder Jack Clark, and Microsoft chief scientist Jaime Teevan each argued, in different settings, that wide exposure to subjects such as the arts, science, engineering, history, and liberal education helps people frame better questions and think more clearly about what they want from AI systems. Musk made the case in a recent interview with China Central Television, where he said a 20-year-old entering an AI transition period should pursue the broadest possible foundational education. He tied that view to the ability to organize questions for robots and AI systems. In the same interview, he said there could be at least 1 billion humanoid robots within 10 years, and possibly in less time. Clark, speaking at a seminar earlier this year, said his literature background turned out to be useful because it taught him history and how people tell stories about the future. Teevan, in an interview with The Wall Street Journal, highlighted adaptability, experimentation, critical thinking, and the willingness to challenge assumptions. The report also notes a tension in that advice: as models get better at understanding natural language, the value of prompt-formatting tricks is falling, and entry-level roles that once helped people practice those skills are also shrinking.

70

Fable 5

2026-07-07 09:02:12

Fable 5 Tops KernelBench-Mega With 18.71x Speedup and the First Single-Launch Megakernel

Fable 5 has taken the top spot in the latest KernelBench-Mega benchmark by delivering an 18.71x speedup on RTX PRO 6000 using a fully hand-written CUDA kernel. The reported result places it well ahead of Claude Opus 4.8 at 14.4x, GPT-5.5 at 4.34x, and Sonnet 5 at 4.0x. The tested workload was 02_kimi_linear_decode, a Kimi-Linear W4A16 mixed decoding task with 4-bit weights and bf16 activations, under a strict setup that allowed only one autonomous session and a three-hour wall-clock limit. What makes the result stand out is not only the score but the implementation: according to the report, Fable 5 is the first model in KernelBench-Mega to produce a true end-to-end megakernel, compressing the inference path into a single GPU kernel launch per decoded token. Anthropic co-founder Jack Clark said the development could mark the beginning of a recursive self-improvement loop, arguing that once models can optimize the low-level systems used to train and run future models, the feedback cycle may accelerate substantially.

420