Where AI Will Change Everything Next

I use AI every single day. For research, for organizing my work, for untangling problems I’d otherwise burn hours on. What I don’t do is let it write my blog posts, paint anything I’d call mine, or compose a song. That’s not a rule I picked at random, it’s a line I think matters, and this post is both my case for it and a look at where I think AI is about to matter most.

My Take: AI Belongs in the Lab, Not the Studio

I think AI might be the single most useful invention humanity has produced in my lifetime, and I don’t say that lightly. Fields that have moved at a crawl for decades, drug development, materials science, infrastructure planning, are suddenly moving in years instead of decades. That’s not hype, it’s already happening, and I’ll walk through exactly where below.

But I think there’s a second, separate question that gets flattened into the same conversation: should AI also write our books, paint our art, and compose our music? My answer is no. Research benefits from speed and scale. Creativity doesn’t work the same way, and I’ll get to why further down.

Where AI Is Already Changing Medicine

Healthcare is the clearest example of AI doing something research alone couldn’t do fast enough on its own. A few things worth knowing:

• Drug discovery timelines are shrinking. Traditional drug discovery typically takes years just to reach an initial filing. AI-driven platforms are now screening and predicting drug candidates computationally before a single wet-lab experiment happens, compressing years of trial and error into a fraction of the time.
• Real trials are already underway. Insilico Medicine’s AI-designed drug for a lung disease (idiopathic pulmonary fibrosis) has reached Phase 2 trials, the furthest any fully AI-designed molecule has gotten so far. No AI-designed drug has FDA approval yet as of mid-2026, so this is progress, not proof.
• Diagnostics and clinical support are already routine. Ambient AI scribes, predictive analytics, and imaging analysis are now standard tools in many hospitals and clinics, not experimental pilots.

Worth being honest about: some scientists remain skeptical that AI-discovered compounds are actually more effective, not just faster to reach trials, and the real test will be whether Phase 3 results hold up. I think that skepticism is healthy, not a reason to dismiss the field.

How AI Is Reshaping Architecture and Infrastructure

This is the area I find most underrated in the AI conversation, because it barely makes headlines compared to chatbots.

• Generative design tools (like Archistar’s add-on for Autodesk Forma) now generate dozens of building layout options in seconds, adjusting for height, zoning, and environmental constraints that used to take architects days to iterate through by hand.
• Digital twins — live virtual models of buildings and infrastructure, fed by real-time sensor data — are becoming standard for tracking energy use, occupancy, and maintenance needs before small problems become expensive ones.
• Cities are shifting from reactive to predictive maintenance. Los Angeles, for example, now continuously monitors water and power infrastructure to catch stress signs before outages happen, instead of fixing things after they break.
• Urban planning simulations (like UrbanSim) model how a city’s infrastructure needs will shift years in advance, based on projected growth and environmental impact.

None of this replaces the architect’s judgment. It replaces the weeks of manual iteration that used to stand between an idea and a testable design.

The Next Frontier: Climate Science and Materials Discovery

If I had to bet on where AI produces the most benefit to humanity over the next decade, this is it.

• Google DeepMind’s materials-discovery project (GNoME) has already predicted the stability of over two million new crystal structures, work that would have taken centuries using traditional lab methods, with real potential applications in batteries, solar cells, and carbon capture.
• Battery research at institutions like Stanford, MIT, and Toyota Research Institute is using AI to accelerate the search for safer, cheaper solid-state battery materials.
• Climate modeling is being reshaped by foundation models like Aurora, which forecast weather, air quality, and ocean behavior at a fraction of the computational cost of older simulation methods.
• Carbon capture research is using AI to screen millions of candidate materials for capturing emissions, a task that used to depend on slow, one-at-a-time lab testing.

The honest caveat here: climate AI models are trained on historical data that’s sparse before the satellite era and thin in the Global South and polar regions, so there are real limits to how far these models generalize into genuinely unprecedented conditions. Worth knowing, not a reason to dismiss the progress.

Why I Draw the Line at Creativity

Everything above is about compressing time on problems that have a correct answer buried somewhere in enough data. Creativity isn’t that. A painting, a song, a blog post, these aren’t puzzles with an optimal solution waiting to be computed faster. Their entire value comes from being made by someone with an actual life behind the work, someone who chose those words or that color or that chord for a reason that came from lived experience, not from pattern-matching across everything ever written before them.

I don’t think this is nostalgia talking. I think it’s the same reason we still care who painted a piece even when a perfect forgery exists, or why a cover song never quite replaces the original. The making is the point, not just the output.

That’s why every blog post here is written by me, mistakes and all, and it’s why I don’t plan on that changing.