🔮 What is left to do
The future of science may be provable, profitable and incomprehensible
OpenAI has recently unveiled an array of mathematical findings generated by an undisclosed advanced model. These discoveries span across number theory, complexity theory, and mathematical physics, numbering 722 manuscripts in 372 categories. The average outcome required the computational prowess equivalent to three hours of ChatGPT Pro.
According to scientist Derya Unutmaz, these results account for 81% of the significant mathematical advancements over the past three years. While most of these findings have been confirmed through the Lean system, they are not all verified. Even if the actual figure is half of this, it remains a staggering quantity. Many problems that have perplexed the finest human minds for decades were resolved within the span of a single afternoon of computational power.
This scenario prompts a thought experiment: envision millions of superintelligent research agents working tirelessly. Formal systems can validate the proofs, but humans will struggle to comprehend the intricate tapestry of machine-generated concepts. This could potentially usher in a new era in human understanding of the world. For the first time in three centuries since the Enlightenment, we can utilize reason and empiricism to grasp what was once inexplicable.
Written by urgent.news from Exponential View's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
