New AI agents can totally run your life—if you let them ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏
October 14, 2026—A stunning look at a year of the sun's coronal chaos. Plus, tough deep-sea crustaceans. And our reporter Mary Randolph gives AI agents run of her life for one week. Scroll below to read how it went.
—Andrea Gawrylewski,
Chief Newsletter Editor
P.S. Know someone who would love this newsletter? Recommend it to them with this link!
|
|
NASA's PUNCH mission stared at the sun for a full year. Check out a video of the chaotic solar corona. | 3 min read
Anthropic claims it discovered a curious stretch of DNA in a virus that repeats over and over again, similarly to CRISPR, the basis of the powerful gene-editing tool. | 7 min read
Earth’s life-support systems are being pushed beyond stable limits—and changes are speeding up, a new report says. | 2 min read
We sat down with Oxford researcher Saul Justin Newman to discuss his new book, which questions the data supporting the existence of "Blue Zones"—places on Earth where people supposedly live to be 100 at astonishingly high rates. | 17 min podcast
When microplastics come into contact with water, the water near their surface can become chemically active, generating highly reactive molecules. This might be how they cause inflammation in the body. | 2 min read
An ancient Chinese math puzzle shows how to win at cards, horse races and gerrymandering. | 11 min read
Leaves change color primarily because of seasonal shifts in sunlight. But extreme weather conditions, such as droughts, heavy rainfall or heat, can affect the timeline of the color change. Here's what to expect this year. | 3 min read
Join the discussion: How have this year's leaf colors been where you live? Share your observations here.
|
|
AI in My Life
How much of your life would you hand over to an AI assistant? Meta and OpenAI are betting it’s a lot. Their new agents, Muse and dots, can read your e-mail and spend your money, and they’re designed to feel like your friends. I spent a week trying out the tools, which I named Mark (à la Zuckerberg) and Sam (à la Altman). The result? Both helpful and terrifying. Though my assistants learned my music taste, anticipated my e-mail habits and could run a mean Facebook Marketplace negotiation, I found myself thinking less and delegating more over the course of the week—sometimes with worse outcomes than if I had just done the tasks myself.
Approachable AI: AI agents are more than chatbots—they have their own “computer” and can complete tasks for you. While the technology isn’t new, Muse and dots make this type of service available to a huge commercial audience with almost no learning curve. Both agents are built to look like a text message chain with a friend: you give them names and cuddly avatars, and their language is intentionally more conversational than a traditional chatbot’s.
What they did: During our week together, Mark and Sam helped me haggle with Facebook Marketplace sellers (I’m now the proud owner of a Muse-sourced dining table); book a U-Haul; organize my budget; and even find upcoming concerts based on my Spotify account. By the end of the week, Mark had the run of my e-mail, bank accounts, social media and Spotify. Cue the shrieks of horror from my friends and family.
What experts say: William Overman, a Ph.D. student who studies AI safety at Stanford University, believes these types of agents are the future of AI and anticipates we may even see physical versions eventually. But he worries that if we assume these assistants are omniscient and increasingly allow them to act on our behalf, it could make it difficult to discern where our work ends and theirs begins. “It’s a further creep into all forms of knowledge work,” he says. “What you’re doing becomes more intertwined with their product.” —Mary Randolph, News Intern
Join the discussion: Have you tried using an AI agent? What role would you be willing to let an agent play in your life? What kind of tasks would you assign it? Click here to share your thoughts.
|
|
A supergiant deep-sea isopod can go for five years without eating. MagicColors/Getty Images
|
|
Tough Crustaceans
In the extreme and sometimes resource-scarce environment of the deep sea, giant armored crustaceans can survive without a meal for more than five years. New research explains how: Millions of years ago, they likely co-opted a gene from bacteria that lets them gorge on a rare meal and slow their metabolism to survive between meals. The researchers sequenced full genomes from different species found at varying ocean depths and compared their energy demands, enzyme activity and oxygen use. They found that deep-sea isopods expressed a key gene called ND1 much more strongly than their shallower counterparts, which helps lower metabolism in cold temperatures.
Why this is interesting: ND1 looks more like a gene found in bacteria than an animal. It lacks filler genetic segments called introns and produces very tiny proteins, both signs of a bacterial gene. The researchers think genes from bacteria living in an isopod’s stomach may have been incorporated into its genome more than 16 million years ago and were passed to future generations.
What the experts say: “This is surprising because bacteria and animals are very different, and such [gene] transfers are rare,” says Jianbo Yuan, a marine biologist at the Chinese Academy of Sciences’ Institute of Oceanology and co-author on the new research paper. —Emma Gometz, Newsletter Editor
|
|
|
|
|
The green and yellow quadrilaterals are squares with areas of X and 16X, respectively. The three other quadrilaterals are rectangles with areas of 272, 300, and 315. What is the perimeter of the rectangle composed of these five areas?
To see the answer to this puzzle, sign up for our math newsletter Proof Positive. Next week's edition will contain the solution.
|
|
This newsletter is powered by caffeine, existential curiosity and subscribers to Scientific American. Join us: 90 days for just $1.
|
|
I have a discrete list in mind of the tasks I would be willing (happy, even) to offload to an AI assistant—making doctors' appointments, paying bills, doing taxes, researching restaurants and making reservations. But even with that list, I am not sure I'd feel safe to turn over my financials or trust the tool to pick the right restaurant. What would you be comfortable with and what conditions would need to be met for you to feel comfortable? Share your thoughts.
|
|
—Andrea Gawrylewski,
Chief Newsletter Editor
|
|
|
|
|
Subscribe to this and all of our newsletters here.
|
|
Scientific American
1000 N West Street, Suite 1200
|
|
|
|
|
