The future of math might be AI-powered

The AI for Math Initiative could reshape how we explore science

Mathematics has always been driven by human insight. Now, Google DeepMind is asking: What if AI could help solve the unsolved?

• DeepMind just launched the “AI for Math Initiative” with five elite research institutions.

• Partners include Imperial College London, the Institute for Advanced Study, IHES, UC Berkeley’s Simons Institute, and TIFR.

• The goal? Use AI not just to compute but to discover, hypothesize, and prove.

• DeepMind’s tools like Gemini Deep Think, AlphaEvolve, and AlphaProof will support the research.

• Their AI has already bested Olympiad problems and set records in algorithm discovery.

• One breakthrough: a new matrix multiplication method that beats a 50-year-old benchmark.

• This isn’t just about math, it’s about redefining the boundary between machine and scientific creativity.

Here’s what we think:

This may be the most ambitious application of AI we’ve seen yet not to replace researchers, but to work alongside them. If it works, AI won’t just help solve equations; it’ll help rethink the very questions we ask in science.

The Magnificent Seven are soaring but cracks are forming

AI hype has pushed tech stocks to record highs. But a closer look shows a market increasingly concentrated and increasingly fragile.

• The tech sector has outperformed in 2025, with AI giants leading the charge.

• The “Magnificent Seven” now make up 36.6% of the S&P 500 up from 12.3% in 2015.

• Nvidia alone has gone from 0.8% to over 21% of their combined value.

• This growth is powered by investor belief and massive AI infrastructure spending.

• But some analysts warn that the S&P 500’s success hides overexposure to a few massive tech stocks.

• Equal-weighted indexes are underperforming, showing gains are not widespread.

• Meanwhile, emerging markets and non-U.S. stocks are quietly outperforming with lower valuations.

Here’s what we think:

We’re watching a market story that’s being written by just a handful of characters. The question isn’t whether AI is real, it’s whether betting the entire index on it is wise. For most investors, diversification isn’t just smart; it’s a hedge against overconfidence dressed up as momentum.

A licensing deal with Getty Images signals a strategic pivot

AI search startup Perplexity has been under fire for scraping and plagiarism. Now it’s taking a different approach by cutting a licensing deal with Getty Images.

• The multi-year agreement lets Perplexity display Getty photos in its AI search tools.

• It formalizes what was previously a quiet partnership under Perplexity’s ad-revenue sharing program.

• Getty’s involvement legitimizes Perplexity’s past use of its images some of which were previously flagged.

• The deal arrives as Perplexity faces a lawsuit from Reddit over allegedly illegal data scraping.

• Both companies are now emphasizing attribution, credits, and linking back to original sources.

• Perplexity argues that its AI use of publisher content falls under “fair use” a claim still hotly debated.

• The deal may set a precedent for how AI startups handle rights and relationships with content owners.

Here’s what we think:

This move isn’t just about legality, it’s about survival. As AI companies scale, they’re hitting walls that tech alone can’t solve: rights, attribution, and trust. Perplexity’s deal with Getty might be one of the first signs that startups are realizing they’ll need real partnerships not just scrapers to compete.

Prohuman team

Free Guides

Explore our free guides and products to get into AI and master it.

All of them are free to access and would stay free for you.

Feeling generous?

You know someone who loves breakthroughs as much as you do.

Share The Prohuman it’s how smart people stay one update ahead.