Academic research - Relatix Health

June 9, 2026

The Download: whole-body rejuvenation drugs and five things to know about AI

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.

David Sinclair plans to test whole-body rejuvenation drugs in the XPrize competition

The outspoken longevity scientist David Sinclair has predicted that, one day, you’ll go to the doctor and get a prescription that will make you 10 years younger. MIT Technology Review has learned of his latest step toward this: human tests of a “reprogramming” drug.

Sinclair, a biologist at Harvard Medical School, plans to launch the tests in a $101 million competition organized by the XPrize Foundation. The winners will “restore” a person to an earlier apparent age, as measured by improvements in immune, cognitive, and muscle function.

The grand prize goes to any team able to show a 10-year (or greater) relative improvement after one year of treatment.

Sinclair says he plans to give an oral drug mixture to volunteers, in a bid to seek “evidence for age restoration in humans.” Find out how he hopes to reverse ageing through chemical reprogramming.

—Antonio Regalado

Five things you need to know about AI

—Will Douglas Heaven

At SXSW London last week, I gave a talk called “Five things you need to know about AI,” in which I shared what I think are the biggest themes in AI right now.

I pulled a few things from our first AI10 list, an annual guide to the top trends in this buzzy world, but I also veered off on several tangents. In my half-hour slot, I tried to cover the key talking points that I think help to make sense of what’s going on in tech—and thus the economy—today.

Five key thoughts emerged: AI is everywhere all at once, it’s getting scary, a backlash is growing, it’s becoming a big deal for science—and I didn’t even need to show up at the talk. Read the full story for all the details.

The must-reads

I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.

1 OpenAI has confidentially filed for a US IPO
The listing could come as early as September. (Reuters $)
+ OpenAI is targeting a valuation of up to $1 trillion. (Financial Times $)
+ The IPO will test investor appetite for AI companies. (WSJ $)
+ The move follows IPO filings from Anthropic and SpaceX. (CNN)

2 The US claims BYD, Baidu, Alibaba, and others are aiding China’s military
The Pentagon added them to a list of military-linked companies. (WSJ $)
+ The designations limit their operations in the US. (BBC)
+ The new additions also include humanoid firm Unitree. (TechCrunch)
+ The Pentagon is adapting to China’s tech rise. (MIT Technology Review)

3 Apple’s long-awaited AI overhaul of Siri is finally here
“Siri AI” promises to be a more conversational assistant. (NYT $)
+ It includes a standalone app and screen-reading features. (Reuters $)
+ And arrives after two years of repeated delays. (Axios)

4 The White House and Congress are working to limit state AI laws
A new deal would curb state rules for federal legislation. (Axios)
+ AI regulation has divided US politicians. (MIT Technology Review)

5 Meta is launching a “workforce academy” for building data centers
The five-week program is free of charge and guarantees a job. (WSJ $)
+ It arrives shortly after Meta laid off 8,000 employees. (NPR)

6 Taiwan is mulling curbs on AI chip exports to China
The new controls would further align with US restrictions. (Bloomberg $)
+ Future AI chips could be built on glass. (MIT Technology Review)

7 Meta has quietly removed face-recognition code from its smart glasses app
The code identified by investigators has disappeared. (Wired $)

8 Humanoid robots are edging towards the battlefield
American and Chinese militaries are pursuing the tech. (BBC)

9 The world’s first wind-powered underwater data center has launched
It uses less power and water than land-based equivalents. (Guardian)

10 You could get some benefits of sleep without having to nod off
If new brain stimulation works as well on humans as on mice, that is. (New Scientist $)

Quote of the day

“You’re on the train, but you know that there’s no destination.”

—Clara Shih, a former top AI executive at Salesforce and Meta, tells the New York Times that AI training can’t keep up with the field’s advances.

One More Thing

Inside the race to make human sex cells in the lab

An embryo forms when sperm meets egg. But what if we could start with other cells—if a blood sample or skin biopsy could be transformed into “artificial” sperm and eggs? What if those were all you needed to make a baby?

That’s the promise of a radical approach to reproduction. Scientists have already created artificial eggs and sperm from mouse cells and used them to create mouse pups. Artificial human sex cells are next.

The advances could herald the end of infertility, but they raise major scientific and ethical challenges.

Read the full story on the new recipes for sperm and eggs.

—Jessica Hamzelou

We can still have nice things

A place for comfort, fun, and distraction to brighten up your day. (Got any ideas? Drop me a line.)

+ These chefs turn Pop-Tarts into the desserts that inspired them.
+ A choir has beautifully transformed System of a Down’s “Chop Suey!”
+ Scientists finally traced crabs’ sideways walk in this fascinating study of evolution.
+ This nostalgic essay on the family computer is a touching throwback to early internet life.

Top image credit: Stephanie Arnett/MIT Technology Review | Getty Images

Please send Pop-Tarts to hi@technologyreview.com.

You can follow me on LinkedIn. Thanks for reading!

—Thomas

June 9, 2026

Trajectories and predictors of self-care from pre-discharge to 12 months after discharge in people with spinal cord injury: a longitudinal study using growth mixture modeling

IntroductionSelf-care is essential for preventing secondary complications and supporting community reintegration after spinal cord injury (SCI). However, evidence is limited on how self-care changes over time after discharge and whether distinct subgroups exhibit different trajectories.MethodsUsing convenience sampling, 220 patients with SCI were recruited from a tertiary hospital in northern China between August 2023 and December 2024. Self-care was measured with the Self-Care in Spinal Cord Injuries Inventory (SC-SCII) at pre-discharge baseline and at 1, 3, 6, and 12 months after discharge. Baseline mental health literacy and perceived social support were assessed using the Multicomponent Mental Health Literacy Scale (MMHL) and the Perceived Social Support Scale (PSSS). Repeated-measures ANOVA was used to describe the population-average time trend, and growth mixture modeling (GMM) was used as the primary person-centered analysis to identify latent trajectory classes. Multinomial logistic regression examined predictors of class membership.ResultsA total of 209 patients completed all assessments (attrition rate: 5.0%). Mean SC-SCII scores peaked at 1 month post-discharge and then declined gradually over 12 months (F = 25.965, p < 0.001). GMM identified three distinct self-care trajectories: low-level decreasing (31.1%), moderate-level stable (39.7%), and high-level increasing (29.2%), with high classification accuracy (entropy = 0.965). Sex, educational level, mental health literacy, and perceived social support were associated with trajectory membership.ConclusionSelf-care trajectories after SCI are heterogeneous, and early post-discharge improvements may be transient for many individuals. The identified trajectories provide preliminary evidence for developing future prediction tools and trajectory-informed transitional nursing interventions.

June 9, 2026

Pulvinar and total thalamus volumes are preserved following early monocular enucleation

BackgroundMonocular enucleation, the surgical removal of one eye, occurs early in life and leads to changes in visual, auditory, and audiovisual processing in adulthood. These changes can be observed behaviorally, as well as through cortical structure and white matter connectivity of visual and auditory pathways. Subcortically, the thalamus is a critical sensory processing structure that modulates both unisensory and multisensory stimuli, which are later processed in the cortex. Previous studies have shown that following monocular enucleation, the lateral geniculate nucleus (LGN) is reduced in volume, although this reduction is less than predicted. In contrast, the medial geniculate body (MGB) is asymmetric but maintains its volume. Together, this may support the auditory and audiovisual enhancements observed following early monocular enucleation. Another key subcortical thalamic nucleus, the pulvinar, plays a broad role in human visual information processing and sensorimotor integration.MethodsThe current study used structural MRI to anatomically localize and measure the total pulvinar and its subnuclei, as well as total thalamus volume, in individuals who had undergone early monocular enucleation during postnatal maturation compared to binocularly intact controls.ResultsOverall, people with one eye demonstrated preserved pulvinar and total thalamus volumes compared to binocularly intact controls.ConclusionThe preserved structural volume of the pulvinar and total thalamus may support the intact lower-level auditory and audiovisual processing previously observed in individuals with one eye. The absence of pulvinar volume changes in this broad-function supporting, subcortical region builds on previous studies regarding thalamic plasticity after early monocular enucleation. These findings provide evidence that not all thalamic nuclei show measurable long-term volumetric alterations and that neural plasticity is both regionally and functionally dependent.

June 9, 2026

Overlaps and differences in the core symptoms of patients with attention-deficit/hyperactivity disorder and patients with borderline personality disorder

BackgroundIndividuals with attention-deficit/hyperactivity disorder (ADHD) and individuals with borderline personality disorder (BPD) show symptomatic overlaps. They both suffer from deficits in emotional regulation, are impulsive and have problems with their self-concept. Therefore, a precise diagnostic differentiation is of great importance. The aim of this study was to find symptom overlaps and differences in patients with ADHD and BPD.Methods80 patients with ADHD, 55 patients with BPD and 55 healthy controls were examined regarding their ADHD and BPD symptoms and their degree of emotional dysregulation using self-report instruments.ResultsPatients with ADHD and patients with BPD did not differ significantly in their expression of emotional dysregulation. However, the ADHD patients showed higher scores in impulsivity, inattention, and hyperactivity, whereas the group with BPD showed higher scores in self-concept problems and suicidal behaviour. The two clinical groups showed significantly higher scores in emotional dysregulation and all other symptom domains compared to the control group.ConclusionThe symptom overlap in emotional dysregulation yields implications for both further research and diagnosis of ADHD. Further studies should define emotional dysregulation consistently to examine the same construct. Key Practitioner Message: This article yields implications that individuals with ADHD and BPD have several symptom overlaps and in fact have no difference in their emotional dysregulation. This has a vast importance for differential diagnosis and treatment of ADHD.

June 9, 2026

Cognitive and neuropsychological correlates of the attention training technique: a systematic review and evidence synthesis

IntroductionThe Attention Training Technique (ATT) is a brief metacognitive intervention recognised as a possibly efficacious standalone transdiagnostic treatment for emotional disorders. The cognitive and neuropsychological mechanisms underlying its clinical effects are of particular interest in understanding and developing the technique. The aim of the systematic review was to synthesise and evaluate the cognitive-attentional task performance and neurocognitive correlates of ATT in the context of theoretical mechanisms from which ATT is derived.MethodsFive electronic databases (PsycINFO, MEDLINE, PubMed, Web of Science and EMBASE) were searched from January 1990 to November 2025. Studies that used ATT as part of a metacognitive multi-component treatment package or combined with other therapy/technique(s) were excluded. Sample inclusion was diverse to capture effects on non-clinical and clinical individuals and across age groups for potential sub-group analyses.ResultsIn total, 20 studies with 1, 230 participants met the inclusion criteria. Four studies included clinical samples, four studies included non-clinical participants, two studies used experimental induction of pain or mind wandering, and 10 used healthy samples of which two used school children. Study quality varied from strong to weak with the majority receiving ‘moderate’ ratings. Across 14 cognitive-attentional tasks and three neural methodologies (EEG, fNIRS, fMRI), the review found small to large cognitive and neural effects associated with ATT. Nine cognitive tasks showed significant ATT-dependent effects in at least one study, with the most consistency shown on the emotional dot-probe. Neural findings across all methodologies converged, suggesting that ATT modulates cognitive control, frontoparietal, dorsal attention networks and reduces default mode network connectivity.DiscussionInterpretation and synthesis of findings based on the S-REF model are consistent with cognitive and neural effects involving reduced threat monitoring, improved executive control, and enhanced disengagement from self-referential processing; central theoretical mechanisms and design parameters of ATT. Where inconsistencies across study effects emerged, they may be due to heterogeneity in cognitive task and measurement factors and ATT protocol deviations. Future research on individual differences in neurocognitive effects associated with ATT across clinical and sub-clinical populations is needed. Studies must safeguard fidelity and adherence to the ATT protocol and improve reporting of these important factors.Systematic review registrationhttps://www.crd.york.ac.uk/prospero/, identifier CRD42024483053.

June 9, 2026

Optimizing Digital Cardiac Rehabilitation Using the Multiphase Optimization Strategy: Mixed Methods Feasibility Study

Background: Cardiac rehabilitation (CR) is an evidence-based, multicomponent intervention. However, participation in and reach of CR remain suboptimal globally. Digital CR is a promising alternative to traditional center-based CR, with the potential to increase intervention reach and efficiency. However, efforts to increase the efficiency of digital CR require an understanding of the relative effectiveness of the components of CR, which is currently lacking. The Multiphase Optimization Strategy provides a framework to evaluate the effects of individual components within complex interventions. Objective: This mixed methods study explored the feasibility and acceptability of implementing the procedures of a factorial design and delivering multiple intervention components in preparation for an optimization randomized controlled trial of a digital CR intervention. Methods: Patients attending CR in a community setting were randomized to 1 of 8 experimental conditions in a 2 × 2 × 2 (23) factorial trial design. Each condition received a different combination of three intervention components over a 6-week study period (1) goal setting and self-monitoring, (2) education, and (3) feedback messages. Feasibility was assessed through intervention fidelity (eg, usage statistics) and outcome measure data completeness. Acceptability was measured using the System Usability Scale, a questionnaire, and semistructured interviews based on the Theoretical Framework of Acceptability. Results: A total of 8 participants were recruited and retained in the study. The mean age was 75 (SD 5.6) years, and the majority were female (5/8, 62.5%). The digital CR intervention demonstrated good usability (System Usability Scale score 72.1, SD 19.1), and 83.3% (5/6) of participants found the digital technology acceptable. However, only half (2/4, 50%) found the feedback messages acceptable. Fidelity was high for goal setting/self-monitoring and feedback but lower for education. Qualitative findings indicated that participants held positive attitudes toward the intervention and reported improvements in physical activity, although many expressed a preference for more tailored feedback and 2-way communication. Of the 3 prespecified progression criteria, usability met the “Go” criterion, whereas intervention fidelity, acceptability, and outcome measure data completeness met the “Amend” threshold. Conclusions: This study demonstrated the feasibility of implementing a factorial design and delivering multiple intervention components within a digital CR intervention. While the intervention was generally acceptable, modifications to the education and feedback components are necessary prior to conducting a pilot optimization randomized controlled trial.

June 9, 2026

Learning to lead in a hybrid human-AI enterprise

As adoption of AI agents looks set to surge by as much as 300% in the next two years, leadership teams are carefully considering the implications of a hybrid human-AI workforce.

Unlike existing enterprise-level automation that relies on manual input, AI agents are capable of autonomously coordinating complex tasks, interacting with multiple tools and environments across an organization. In early applications that center on customer service, HR, and sales, adoption of agentic AI has led to productivity gains of 30-50%.

Their autonomy positions agents more as collaborators than tools, working side-by-side with human employees in blended teams that look poised to upend traditional workplace dynamics.

More than three-quarters of HR leaders believe that the deployment of AI agents will transform existing workplace norms, driving a complete reappraisal of how roles and responsibilities are distributed, how skills are prioritized, and how workplace culture is shaped.

Though many admit they’re in the early or preparatory phase of this shift, 86% of chief HR officers predict that navigating digital labor shaped by agentic AI will be a central component of their role in the years ahead.

Fluency in the change management aspect of agentic AI adoption will be a crucial differentiator when it comes to unlocking the full potential of the technology going forward, believes Ateet Jayaswal, chief culture and employee experience officer at Wipro, a leading technology services and consulting company. This moment is one that he says, “calls for a mindset shift in how HR leaders would enable their organizations.”

Redeploying roles to enable higher-value work

As AI agents assume ownership of more complex and integral tasks, the distribution of roles and responsibilities within an organization will undergo significant change. It’s estimated that three-quarters of current roles will require redesign, reskilling, or redeployment by 2030 as a result of agentic AI.

For leadership, this shift should be about reskilling employees toward higher-value work in order to optimize the potential of an agent-human hybrid workforce, says Jayaswal.

For example, Wipro is a complex organization of 240,000 employees across 65 countries. It previously had multiple policies, documents, and knowledge fragmented across different systems, which delayed response to employee queries.

But the company has recently integrated a custom agentic AI assistant—an agent co-created in partnership with enterprise agentic AI platform Ema Unlimited—that can swiftly navigate this complex system, assuming responsibility for 50 HR tasks that had previously fallen to human employees. With the help of an AI agent, average response time to queries has lowered from 48 hours to five seconds.

Human employees have more time to focus on work “that requires a creative and imaginative mind and cross-functional collaboration, leveraging diverse ideas and thoughts to problem-solve,” says Jayaswal. The AI agent, meanwhile, handles rote administrative tasks like sorting timesheets or helping employees navigate policies and take actions in the flow of work.

When reallocating employee responsibilities, though, it is imperative that humans remain in the loop, Jayaswal caveats. When agentic AI is incorporated into enterprise technology, it must work with sensitive and personal data and therefore needs even more stringent guardrails and constraints than consumer applications. “When you expose an AI agent to organizational data, when you integrate it into multiple enterprise systems, then pathways around the AI agent become extremely important,” he says. “It’s an evolving space that leadership needs to have front-of-mind.” Governance should include robust data privacy rules and the establishment of governance layers, such as an AI council, he suggests.

At a fundamental level, the adoption of AI agents will force a re-evaluation of human roles, believes Jayaswal. Rather than employees primarily performing repetitive tasks or troubleshooting, a significant proportion of their time will shift to designing, teaching, and optimizing an AI agent that can do this work for them with far greater speed and predictability and without the agent getting bored.

“The nature of your job changes from being the hero who comes in to solve the problem to designing the hero who can solve the problem,” he summarizes. “The individuals who I have seen thrive in this environment are the ones who make this shift.”

An evolving employee skillset

Just as roles and responsibilities will be reconfigured to reflect the input of AI agents, the core skills of human employees will be reprioritized. More than four in five HR leaders say they’re planning to reskill workers to become more competitive in a market shaped by AI agents.

Technical skills will be increasingly important. Leading employers such as Salesforce, Danone, and Walmart are already rolling out dedicated AI and digital skills programs that aim to equip everyone from frontline workers to C-suite executives with a baseline level of AI literacy in response to the pervasiveness of the technology.

But desirable soft skills will also evolve, Jayaswal points out. Employees who assign tasks to an AI agent need to plainly articulate what modular steps may be needed to accomplish a task, what the desired outcome should be, and what parameters or guardrails need to be in place to ensure the agent doesn’t access or share confidential data.

As HR executives adapt to a blended workforce, three skills are emerging as top priorities during recruitment, according to a recent survey: relationship building, like forging constructive partnerships and account management; collaboration; and adaptability.

Maintaining a healthy workplace culture

In freeing up human employees to focus on higher-value tasks, the hope is that AI agents can elevate the employee experience, deepening fulfilment and satisfaction in the workplace.

“At Wipro, our vision is to improve the life of Wiproites,” says Jayaswal. “We are taking away non-value added work by embracing modern ways of collaborating, engaging, and transacting, leaving associates with higher order work content.”

But leadership teams embracing agentic AI will also need to plan for the new pressures and stressors that the technology can place on a workforce.

There is already confusion and knowledge gaps, with 73% of HR leaders reporting their employees don’t yet understand how digital labor will impact their work. Many organizations have opted to define AI agents as teammates or colleagues on org charts, but new research says this could erode trust and a sense of professional identity. It also raises new questions around accountability and ownership.

The role of management in addressing these concerns is critical, says Jayaswal. To maintain healthy dynamics, managers need to become skilled at orchestrating blended systems, splitting their focus between supervising AI agents and motivating human employees as they also build and supervise AI agents.

Upgrading employee well-being programs will be a core part of maintaining a robust workplace culture. “As there are more interactions with AI agents, you are losing some of the human touch that was provided by service delivery partners or leaders, or often even by colleagues and peers,” Jayaswal says. Employee services that encourage social connection and empathetic communication may help teams navigate this.

A breakneck transformation

Agentic AI looks set to scale at breakneck speed across many enterprises, and it will significantly transform how these organizations operate.

Carefully considering and deciding how to adapt to this newly blended workforce is now a top priority for leadership teams. Reviewing and refining organizational strategies is essential for optimizing both technological gains and the employee experience.

This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff. It was researched, designed, and written by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.

June 9, 2026

Genome Wide Analysis Broadens Genetic Knowledge of Anxiety

A large European meta-analysis led by King’s College London has revealed 39 new areas of the genome linked to symptoms of anxiety and provided new estimates of heritability of this common condition.

Writing in Nature Human Behaviour, the researchers explain they found 80 variants linked to anxiety in 74 specific areas of the genome, but 35 were already known to be linked to the disorder from previous research.

“These correlations highlight the interconnection between mental and physical health,” said Brittany Mitchell, PhD, a senior researcher at the QIMR Berghofer medical research institute and co-first author on the study in a press statement.

“Importantly, while some shared genetic variants may increase risk for both a physical health condition and more severe anxiety symptoms, it’s also true that living with chronic pain or illness can contribute to anxiety symptoms. Our findings don’t reveal causation or the direction of effect, but they do open up important questions for future research.”

Anxiety disorders are very common mental health conditions across the globe with large variations in severity from mild to debilitating. Previous research has attempted to assess the genetic heritability of anxiety but estimates of how heritable this condition is vary widely.

In this study, genetic data from 14 cohorts of people with generalized anxiety symptoms assessed via self-report questionnaires between 2007 and 2023 was combined to carry out a large meta-analysis of 693,869 people. Most of the cohorts were largely made up of people of European ancestry, for example, the UK Biobank, and mostly came from Europe or North America.

In addition to finding the new variants, the team also created a polygenic risk score for anxiety aggregating genome-wide SNP effects. Overall, the estimated SNP-based heritability was around 6% and the team estimated that about 1-3% of variance in anxiety symptom severity could be attributed to genetics according to the polygenic risk score.

Notably the research also highlighted some links between the genetics of anxiety and other conditions like irritable bowel syndrome, coronary artery disease, and migraine.

“Given the high and rising rates of anxiety, especially in young adults, it is more important than ever to improve our ability to identify and understand sources of risk,” write the authors.

“Despite its public health impact, progress in anxiety genetics lags behind other major mental health conditions. We hope our findings encourage genome-wide investigations leveraging existing but potentially underutilized anxiety severity data in genotyped cohorts, accelerating our progress in understanding the genetic architecture of anxiety.”

The post Genome Wide Analysis Broadens Genetic Knowledge of Anxiety appeared first on Inside Precision Medicine.

June 9, 2026

Five things you need to know about AI

At SXSW London last week I gave a talk called “Five things you need to know about AI,” in which I shared what I think are the biggest themes in AI right now.

I pulled a few things from our first AI10 list, an annual guide to the most important trends in this buzzy world, but I also veered off on a number of tangents. In my half-hour slot, I tried to cover the key talking points that I think help to make sense of what’s going on in tech—and thus the economy—today.

(I gave a talk with the same title at SXSW London last year with five different things you needed to know. A lot has happened since then!)

So: This is how I’m thinking about AI midway through 2026. Let me know if you would pick different points!

1. Strictly speaking, I didn’t need to show up to give this talk.

Tongue in cheek? Maybe. But generative AI tools have already become mundane, used by millions to automate everyday office tasks (including producing and delivering talks). It’s no surprise that one of the biggest questions out there right now is what this all means for jobs. People are confused and scared.

The frustrating answer is that despite the hype coming from the top about the potential for AI to join the workforce soon—and viral social media posts yelling that something big is happening—there is almost no data to say either way what kind of effect this technology will have on employment and the economy overall. That’s not to say it won’t have an impact, even a huge one, but it’s just too soon to tell.

In theory, teams of agents working together toward common goals could become assembly lines for white-collar work, doing to offices this century what Henry Ford’s innovations did to factories in the 20th century.

In theory. Because in order to know what will happen to jobs, we need to know what will happen inside the companies that create those jobs. But most companies are still figuring that out.

2. AI is getting scary (for real this time).

There have been scary stories about AI for years—claims that it will kill us all or bring about the end of civilization. There’s still a loud crowd of doomers, but those scenarios remain dystopian science fiction.

What’s happened instead is that many of the worst near-term, real-world fears have come true.

Take deepfakes, AI-generated images or videos of people doing things they didn’t actually do. Deepfakes have been used to incite violence, swing votes, and sow distrust. Trump’s White House is among those creating and publishing fake images.

Many deepfakes are also used to abuse women and girls. One study found that 98% of deepfakes are pornographic and 99% involve women.

Another concern is the rise of dangerous and delusional relationships with chatbots. Many people turn to chatbots to seek private advice and to feel heard. But there are now multiple lawsuits against AI companies alleging that the technology encouraged or aided suicides and other forms of self-harm.

AI is also being used in warfare in new and worrying ways. LLMs are now giving advice, not just being used for analysis. One US defense official told my colleague James O’Donnell that you could now give a military chatbot a list of targets and ask which one to hit first. Anyone who uses AI knows that its output needs to be reviewed carefully. In fact-paced, high-stress active conflict, the risk that corners get cut is high.

3. A lot of people really hate AI.

I checked out an anti-AI protest in London earlier this year and found a very broad mix of complaints. Banners proclaiming the end times bounced along to chants of “Stop the slop! Stop the slop!” Protests are getting more organized and drawing larger crowds.

There’s pushback from fans of films and video games, who object to the use of generative AI in their favorite titles. In one notable case, the acclaimed 2025 game Clair Obscur was stripped of an award when the developers admitted to using AI in just one small, specific part of its production.

And there’s the data center backlash. The US has more than 5,400 data centers and counting. With the energy demands of AI growing, people are unhappy about the environmental impact and their rising electricity bills. Activists are managing to stall development in a number of places.

Regulation is becoming politically popular. Grassroots movements like QuitGPT have gained momentum. A small number have turned to violence; a few weeks ago somebody threw a Molotov cocktail at Sam Altman’s house. It’s not clear where all this leads. But the apocalyptic hype from tech leaders is not helping people stay calm.

4. AI for science is a very big deal.

It’s early days yet, but the potential for AI to help make a genuine and important scientific discovery is greater than ever.

Google DeepMind has developed Co-Scientist, a multipurpose tool that can help researchers dig up and compare previous results, generate hypotheses, and devise experiments to test them. OpenAI told me this year that its North Star is the goal of building a fully automated researcher by 2028.

Mathematicians are excited too. Fundamental math underpins many everyday technologies, from internet security to video streaming. The last few months have seen a string of claims that AI has cracked unsolved math problems. And software that can solve really hard math problems will be able—so the argument goes—to solve more general-purpose real-world problems too.

What are the downsides? Some scientists are warning that an overreliance on AI tools could narrow the scope of research because scientists may choose problems that are most suited to AI assistance. There are also concerns that AI-assisted research will lead to a flood of inaccurate or fake results: science slop.

5. AI is everywhere all at once.

So where does that leave us? There are a lot of exciting things, a lot of worrying things, and a lot of hot air. It can be exhausting to keep up, and yet it all feels inescapable. Some people will tell you we’re in a race to the top; some will tell you we’re in a race to the bottom. But it’s really not clear where we’re headed.

AI companies want us to march to their tune and buy into the propaganda about artificial general intelligence, whatever that means. They are selling a vision that feels inevitable, but it isn’t.

We’ve built a technology that can do humanlike things, and I think that makes it hard to get our heads around the fact that it is still just a technology.

Something is happening. Maybe even something comparable to the invention of electricity or the internet. But technologies like that take time to settle and bring lasting change.

Get ready for a marathon, not a sprint.

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here.

June 9, 2026

Opinion: Ending animal testing could set back xenotransplantation just as the field is poised for a breakthrough

Health secretary Robert F. Kennedy Jr. seeks to end all federally funded animal testing after concluding that “the predictivity of animal models is very, very poor for human health outcomes.”

In November 2025, Centers for Disease Control and Prevention staff were told that the agency would be required to phase out primate studies, and they are in the process of transferring their animals to a primate sanctuary. The National Institutes of Health, the largest funder of biological sciences in the U.S., has stopped issuing funding opportunities exclusively for animal models, sending a clear message that basic science conducted in animals small and large will no longer be a priority. Of the eight NIH-funded National Primate Research Centers in the U.S., one has been shuttered, and a second is exploring the possibility of converting to an animal sanctuary.

Read the rest…