48. 对话前 OpenAI 科学家：GPT-5 能获得奥赛金牌，但那可能具有欺骗性
卫诗婕｜漫谈Light the Star
2025/08/23
48. 对话前 OpenAI 科学家：GPT-5 能获得奥赛金牌，但那可能具有欺骗性
48. 对话前 OpenAI 科学家：GPT-5 能获得奥赛金牌，但那可能具有欺骗性
卫诗婕｜漫谈Light the Star
2025/08/23
In this episode, former OpenAI scientists Kenneth Stanley and Joel Lehman reflect on the evolution of AI innovation, the founding principles of OpenAI, and the unexpected rise of ChatGPT. Their earlier work, which emphasized open-ended exploration over rigid goal-setting, has gained renewed relevance in light of recent developments. The conversation also touches on the shift in OpenAI’s culture, the implications of GPT-5’s underwhelming reception, and what the future might hold for AI research.
Stanley and Lehman discuss how innovation often arises from unstructured exploration rather than targeted objectives, a theory validated by the surprise success of ChatGPT. They recount their time at OpenAI during its more experimental phase and express nostalgia for the playful, curiosity-driven research environment. With the release of GPT-5, the broader AI community has voiced skepticism over its advancements, leading to concerns about stagnation. However, they suggest that slower progress could create space for more imaginative research. The conversation also addresses the challenges of sustaining open-ended innovation in corporate environments, the pitfalls of benchmark-driven development, and the importance of embracing unpredictability in scientific discovery.
00:09
00:09
Kenneth Stanley and Joel Lehman left OpenAI in 2023, advocating for novelty-driven research.
03:20
03:20
ChatGPT represents a real-world example of unexpected innovation with massive impact.
08:39
08:39
Novelty search algorithm was created based on these insights about diverse stepping stones
09:57
09:57
Novelty search algorithm outperforms traditional methods in robot simulations
11:00
11:00
Artists emotionally connect with message on innovation
12:30
12:30
Guests describe joining Uber AI Labs and OpenAI as a happy coincidence
15:00
15:00
Sam publicly admired their book
15:49
15:49
Sam Altman described as pragmatic and effective leader during 2020-2022
16:37
16:37
The release of GPT-3.5 triggered fierce competition in the global AI industry
17:36
17:36
OpenAI faces a choice between non-profit exploration and becoming a tech giant
20:47
20:47
One speaker acknowledges feeling a bit of regret about the changes in OpenAI.
22:30
22:30
Preference for basic research over scaling
22:58
22:58
OpenAI succeeded early on by pursuing opportunities without a fixed business plan.
24:00
24:00
The launch of GPT-5 has sparked criticism for squeezing out innovation in the AI field.
24:23
24:23
The speaker questions if scaling alone will lead to AGI.
26:02
26:02
The speaker hopes the current AI paradigm is running out of steam.
26:48
26:48
The speaker doubts Transformer is the final architecture.
27:13
27:13
A big success may lead innovators to focus too narrowly, losing their innovative edge.
28:52
28:52
Google didn't invent the Transformer architecture that sparked the language model revolution.
30:38
30:38
Mature enterprises like Google can avoid stifling innovation by letting go of objectives sometimes.
31:21
31:21
Doing non-useful things can be best for innovation
33:55
33:55
Open-ended exploration in AI research faces different challenges in corporate environments due to profit pressures
35:42
35:42
Success can enable continued exploration and open-ended innovation
35:58
35:58
OpenAI demonstrated high conviction in scaling neural networks despite early models being 'garbage'
38:20
38:20
A computer system achieving a gold medal in the Math Olympiad from a benchmark perspective.
39:24
39:24
New models always perform better on benchmarks, making them seem manipulable.
40:11
40:11
Playing to benchmarks rather than intelligence can be misleading
40:26
40:26
Lila Sciences aims to create scientific superintelligence to drive a new scientific revolution
42:38
42:38
The speaker is surprised by the development of coding models as a significant practical advancement in AI.
44:34
44:34
Debugging unwritten code takes more time, but bugs are expected to reduce, leading to more acceleration.
45:10
45:10
Deepseek's innovation has caught Silicon Valley's attention and changed the dynamics of the tech competition between China and the US
47:10
47:10
Planning may hinder disruptive innovation, and test-based education may not reflect real-world creativity.
