Musk Claims AI Was Already Outdated Two Years Ago, Will Explode in Two Years, Directly Refuting Claims of Slowed Progress and Long-term Profitability Issues
Musk cited former OpenAI employee Will Depue's views and expressed support.
Depue responded to the bearish outlook on frontier lab valuations, stating that while it makes sense to overlook AGI, AI capabilities will not slow down, and recursive self-improvement will arrive within 24 months.
Musk stated that AI was already outdated two years ago and could be put in a museum, and that in two years AI will explode in capabilities, directly refuting claims of slowed progress and long-term profitability issues.
Source: Public Information
ABAB AI Insight
Will Depue was involved in the post-training of OpenAI's Sora and o series, and has previously discussed data bottlenecks and the essence of learning algorithms, emphasizing that AGI is about algorithms rather than weights.
Musk is continuously enhancing computational power and synthetic data pathways through xAI, resonating with Depue's RSI timeline, as funds and talent shift from traditional labs to entities with self-improvement feedback loops.
Similar to the transition of deep learning from academia to industry in the 2010s, we are currently at a critical stage of moving from scale expansion to recursive acceleration.
This essentially represents a technological substitution; once recursive self-improvement is realized, capability growth will shift from linear to exponential, rendering traditional valuation models ineffective as the pace of human R&D is replaced by algorithmic feedback loops.
ABAB News · Cognitive Laws
The miracle of two years ago is a museum piece
The capability curve is determined by recursion rather than linearity
Ignoring the singularity will ultimately become fuel for the bears