Our way of life is going to change faster than you can imagine

In recent years, we have been inundated with news of AI breakthroughs across diverse fields such as medicine, engineering, psychology, education, music, video production, and pharmacology. The pace of these advancements is difficult to comprehend, and the next five years promise to bring profound changes to our lives.

AI has the potential to revolutionize education by enabling students to learn at an accelerated pace. Studies have demonstrated that with AI, students can learn the material in just six weeks, which would traditionally take two years for human teachers to teach. With the advent of level 5 autonomous driving, we can anticipate a 94% reduction in auto accidents. This will dramatically decrease car insurance costs and a significant reduction in emergency room visits due to auto accidents—from approximately 3.4 million visits annually to 204,000. The implementation of Vehicle-to-Everything (V2X) AI technology is expected to further reduce the number of accidents to just 68,000 per year.

In the medical field, AI is poised to replace radiologists and pathologists. Some states are already considering the removal of a human review of AI diagnostics, which will further reduce medical costs. The process of purchasing auto and life insurance will also be transformed, eliminating the need for insurance agents to find the right policy. Legislative changes may be required to accommodate these advancements.

Traffic management will see improvements with AI-based timing of traffic lights, reducing waiting times at intersections when there is no cross-traffic. Autonomous vehicles will be able to travel closer together, increasing road capacity. Additionally, AI software now offers automated telephone receptionist services that speak in clear, natural voices, record and transcribe conversations, and schedule appointments. These services are available for just $10 an hour, a fraction of the cost of hiring a traditional answering service.

These examples are merely a glimpse of the changes we will experience in the near future. The extent of these transformations will be determined by the level of intelligence achieved by AI platforms. Similar to how SAT scores gauge the readiness of high school students for college, a method is needed to assess the intelligence of AI systems. Initial AI system tests are now easily passed by existing platforms with 90% accuracy.

Humanity’s Last Exam

To address the limitations of current benchmarks, the Center for AI Safety (CAIS) and Scale AI have developed a new, advanced benchmark called “Humanity’s Last Exam” (HLE). This comprehensive benchmark tests deep reasoning skills and expert knowledge. To appreciate the sophistication of HLE, it is essential to understand its creation.

HLE comprises 3,000 questions spanning over 100 subjects, including mathematics, humanities, languages, chemistry, and natural sciences. In addition to text-based questions, it incorporates diagrams, images, and other formats. Nearly 1,000 subject matter experts, including professors, researchers, and graduate degree holders from over 500 institutions across 50 countries, contributed questions designed to test the upper limits of AI reasoning and understanding.

Example questions include translating Palmyrene script from a Roman tombstone inscription in the “classics” category and identifying the number of paired tendons supported by a specific sesamoid bone in hummingbirds under “ecology.” Current models like GPT-4 score only 3.3% on HLE, while the best-performing model, DeepSeek-R1 (evaluated only on the text subset), achieved 9.4%.

Classics

Question:

Here is a representation of a Roman inscription, originally found on a tombstone. Provide a translation for the Palmyrene script.
A transliteration of the text is provided: RGYNᵓ BT ḤRY BR ᶜTᵓ ḤBL

Henry T

Merton College, Oxford

Ecology

Question:

Hummingbirds within Apodiformes uniquely have a bilaterally paired oval bone, a sesamoid embedded in the caudolateral portion of the expanded, cruciate aponeurosis of insertion of m. depressor caudae. How many paired tendons are supported by this sesamoid bone? Answer with a number.

Edward V

Massachusetts Institute of Technology