This week in toast, AI has developed free will, just when it starts designing DNA (absolutely nothing concerning here), and America's biggest cities are quietly sinking into the ground. Also, your face can now tell doctors how close you are to death 😬.

AI is all grown up, which is equal measures amazing and terrifying. Frank Martela has proposed that generative AI meets all three philosophical conditions for free will: 

Martela examined AI systems including the Voyager agent in Minecraft and theoretical military drones, finding both qualify as having "functional free will."

This development comes as AI is increasingly deployed in critical areas like healthcare, transportation, and military applications. However, Martela warns that AI has no inherent moral compass, it only has the ethics we explicitly program into it, and as systems gain more autonomy the question of responsibility will likely get even more confusing: are we responsible for the algorithms actions, or are they?

On the plus side, at least all those Philosophy degrees weren’t a total waste of time.

🧐 What's in it for me? As AI becomes more autonomous, products and services will likely improve exponentially, however, we also need to reconsider who's responsible when things go wrong. The morality we program will shape society's future. Or we could keep outsourcing ethics to Silicon Valley. What could possibly go wrong.

💵 Out of the Lab:

Researchers have created an AI that designs synthetic DNA sequences never seen in nature. Great news for biochemists and terrible news for natural selection, which just got shown up by the intern after 3.7 billion years on the job.

The system creates custom genetic instructions: tell AI what you want ("activate this gene in red blood cells but not platelets"), and it determines which DNA letters deliver results. In tests, AI-designed DNA successfully activated fluorescent proteins in mouse cells. 

🧐 What's in it for me? This could revolutionise gene therapy by precisely controlling gene activity only in cells that need treatment. It's like fixing your body's software with targeted patches instead of updating your entire system only to find you’ve wrecked your battery life.

💵 Out of the Lab:

Baby KJ Muldoon has made history as the first person to receive a bespoke CRISPR gene-editing treatment for his unique genetic mutation. Born with CPS1 deficiency that prevents his body from processing proteins, KJ received three doses of custom therapy. 

An international team developed the treatment in just 6 months using base editing to correct KJ's specific mutation. While doctors avoid saying "cure," KJ is thriving with improved protein processing. Meanwhile, it still takes 6 months in the UK just to see a dermatologist.

🧐 What's in it for me? We can now create personalised genetic treatments for one-of-a-kind mutations that most pharmaceutical companies ignore due to the small market size. We're entering a world of tailored genetic medicine.

💵 Out of the Lab:

Researchers have developed an AI algorithm called FaceAge that analyses your face to determine biological age and predict cancer survival chances. Expect boom times for anxiety medications. 

Tests found cancer patients appeared about five years older than their chronological age, with higher FaceAge predictions linked to worse outcomes. FaceAge even outperformed human clinicians at predicting life expectancy. So long to the reassuring "you look great" from your doctor, hello to an AI saying "You look 68.7% dead."

🧐 What's in it for me? This could help doctors make objective treatment decisions based on biological resilience rather than just chronological age, without the need for intrusive tests.

💵 Out of the Lab:

POSTECH researchers have developed a single-layer waveguide that could finally make AR glasses as slim as regular eyewear. This is big news for the multiple tech behemoths that have poured billions into developing AR software, in the expectation that eventually people wouldn’t be embarrassed to wear their tech. 

Traditional AR glasses stack multiple waveguide layers (one each for red, green, and blue light), creating that signature "I'm wearing a computer on my face" look. However, the new design uses "achromatic metagrating" - a sophisticated way of saying we can use 1 layer instead of 3.

🧐 What's in it for me? AR glasses that look normal could make the technology practical for everyday use. Imagine GPS directions, instant translations, or seeing that conference acquaintance's name floating above their head, without a technological brick on your face. 

💵 Out of the Lab:

Boston Dynamics' Spot has a new competitor. Researchers have developed an Amphibious Robotic Dog that navigates both land and water, taking inspiration from how real dogs swim (someone should tell them about fish).

The machine features a paddling mechanism with three distinct gaits: two inspired by the classic doggy paddle for speed, and a trot-like style for stability. Tests showed the doggy paddle reached water speeds of 0.576 km/h, while on land it trots at 1.26 km/h. 

🧐 What's in it for me? More than just a stick retrieval system, amphibious robots could aid environmental research, rescue operations, and military reconnaissance. 

💵 Out of the Lab:

Researchers have defied physics intuition yet again, showing that measurements can be performed on separated quantum particles without bringing them together. The breakthrough hinges on quantum entanglement, that bizarre phenomenon where particles remain connected across any distance as if joined by an invisible string.

Imagine checking what's in two different sealed boxes simultaneously, without opening either one. This capability fundamentally changes how we can process quantum information, potentially solving major challenges in building practical quantum computers where qubits are notoriously finicky about being moved around.

🧐 What's in it for me? This could enable more secure quantum communication networks and practical distributed quantum computing. Complex calculations across multiple quantum computers without physical data transfer could be possible.

💵 Out of the Lab:

All of the Biggest U.S. Cities Are Sinking 🌊

Satellite data reveals in 25 cities, at least two-thirds of urban areas are subsiding. Concrete evidence that things really are going downhill everywhere.

Houston is sinking fastest, with 40% of its area dropping over 5mm yearly and 12% sinking twice as fast. The culprit? We're drinking our foundations away, with groundwater extraction causing 80% of the subsidence.

San Antonio tops the danger list with 1 in 45 buildings at serious risk from uneven ground movement. "This subsidence can produce stresses on infrastructure that will go past their safety limit," warns lead author Leonard Ohenhen.

On the bright side, penthouses may soon get ground-floor convenience at no additional cost.

Until next time, stay curious.

Like what you're reading? Share toast with a friend.