AI's self-improvement problem and what is driving the heat

Researchers find that AI systems still cannot conduct open-ended research, while the El Niño phenomenon intensifies heat waves in the northern hemisphere.

Por El Medio Oriente
24 de agosto de 2026
A three-dimensional scientific visualization of an egg-shaped structure with gray, white, green, and magenta coloring against a black background.
Scientific visualization of a cell or biological structure with different components highlighted in colours: magenta for the external membrane, grey and white for internal structures, and green for specific components. (MIT Technology Review)
1 min de lectura
Tamaño del texto

The boldest promise from the AI industry right now is that AI will soon improve itself, with almost no need for human oversight. However, a new study suggests it could take longer to achieve this.

Researchers found that AI agents still cannot conduct open-ended AI research—that is, research in free form without clear answers that requires the judgment and creativity needed to make genuine advances. The big question now is how crucial open-ended research is for recursive self-improvement, and whether AI systems can get there simply by improving at more limited tasks. The results may moderate claims that recursive self-improvement is within reach.

This summer has been extremely hot across much of the northern hemisphere. June and July marked the warmest two-month period in Europe since records began, the central United States experienced its hottest month ever in July, and South Korea recorded its highest temperature on record.

Climate change makes heat waves more likely and intense. However, there is another factor at play: El Niño, which is already intensifying and is expected to have a greater effect on global temperatures next year. This is the behind-the-scenes story of this summer's extreme heat and why 2027 could be even hotter.

AI's self-improvement problem and extreme heat | El Medio Oriente