GPT‑6 Astra allowed Parallel’s agents to research and synthesize labor-market data in half the time and at half the cost vs. prior models.

How often do you get to talk to a guest who has both an Academy Award and who invented textbook machine learning algorithms? John Platt has an Oscar, two textbook algorithms, two named asteroids, and an Erdos-Bacon number of 6. This was easily the most fun bio of all the guests we’ve read to date. And the result was an epic and fun chat covering Google’s Empirical Research Assistance (ERA), how AI can help battle climate change, and tons of great stories about the co-evolution of science and AI.

John’s colleague Dave Bacon likes to tease John that his career has been defined by being twenty years early to the next big thing. This may be convolutional neural networks (some credit him with coining the term), fusion research, quantum computing. John and Google have been working on solving some of humanity’s hardest problems with AI and computation for well over a decade now. Recently John and his team set their sights on using AI to solve any scientific problem that can be written down as a score.

Google’s Empirical Research Assistance (ERA)

John’s team has taken on many hard scientific problems over the years. In solving these, they noticed a pattern, many scientific problems can be reduced to what John calls a “scoreable task”. Once you have the score function, the goal is to find some code that maximizes the score. The hard part is in formulating the score, but once you have the score finding the maximizer can still be quite a lot of effort.

John’s team set out to automate solutions to this general problem. This came out of the idea of an “auto-Kaggle” AI, which can solve any Kaggle problem you can throw at it. Kaggle is owned by Google, so all the data was ready and easily available to them!

The result is Google’s Empirical Research Assistance or ERA (paper, github, blog).1 ERA is surprisingly simple conceptually. Gemini (or your LLM of choice) keeps a running tree of past experiments (notebooks) and where they’re going. It’s a close cousin of Monte Carlo Tree Search: at each iteration the Upper Confidence Bound rule picks which notebooks are most promising to mutate. This is optimistic, not greedy, so sometimes even the fifth-best notebook gets chosen. Gemini then proposes mutations for each one, about ten at a time. The history of each branch is shared, so different leaves can learn from each other.

“It’s almost like having a hyper-eager grad student who doesn’t sleep.”

Evolutionary algorithms have been around since the 70s, but this works because Gemini actually knows where to look! What’s even more interesting is that there was a step change between Gemini 2.0 and 2.5, and this went from just not working to working great.

ERA is so powerful that John and his team solved many outstanding problems with it, resulting in at least ten papers. Some of these were climate change related, which we talk about in the next section.

So, we had to ask: if you have an optimization god how do you avoid fooling yourself? John’s answer is that ERA provides predictive models. It’s up to the scientist to make sure they’re truly descriptive. Some of this just involves good old-fashioned careful machine learning science. “It’s a power tool. It can slice your fingers off.” This led to some fun discussion about Kaggle competitions, and the fun ways people can overfit to datasets without meaningfully solving the problem you actually care about: Google’s contrail-detection competition was won by entrants who noticed a half-pixel error in the labels (is the origin at the corner of the pixel or the center?) and this turned out to be a part of the winning special sauce. Great for winning $15,000, not so helpful if you actually want to solve contrails.

“People themselves will act like these LLMs and try to reward hack. It goes back to Goodhart’s law: any metric that becomes a target is no longer good as a metric.”

His advice for where to start instead?

“Always just fit linear regression. Just do it. Just do it. Just do it. Or SVM.”

Tackling Climate Change with AI

John and his team have worked extensively to mitigate the effects of climate change. We talked about several of their initiatives.

Perhaps the most interesting result we talked about was reducing the effects of condensation trails (contrails) from airplanes. Those little streaks you see running behind planes somehow account for 1% of all human-induced global warming?!? Some of these trails of ice crystals can hang out for days. These crystals are black in the infrared, acting like a thermal blanket that traps heat day and night.

It’s easy to understand what’s happening here, a region of atmosphere becomes “ice supersaturated”,2 and a tiny bit of exhaust seeds water vapor that instantly crystallizes. The scale here is astounding, with a single gram of exhaust resulting in ten kilograms of ice crystals.

The solution to all of this is quite simple, in principle! We know what parts of the atmosphere are most likely for the trails to form. Just have the planes drop a flight level or two. Problem solved, right? Well, the hard part is accounting for how much warming was prevented. This is a counterfactual problem, parts of which stumped John’s team for over two years. They had a working model for the heat-trapping half, but not for the reflected sunlight. ERA was able to find a simple model with some confounders they hadn’t considered. Cracked it!

Modeling climate generally is a hard problem. Climate is best thought of an attractor of many different possible weather outcomes.3 This makes it much harder to model.

“Weather is where you are on the attractor, and climate is the statistics of the attractor. The problem with climate is that we’re altering it. The attractor itself is changing, it’s moving.”

John and his team have worked on treating both the symptoms and the disease of climate change, with several other works in the area. Another fun example we briefly cover is FireSat, a way of using a constellation of satellites to rapidly identify fires before they grow too big to put out. For anyone living in California, you understand the problem. In dry years a small fire can result in hundreds of thousands of acres. If you could find this fire when it’s the size of a room, it could be put out. By the time it hits an acre we have a much harder problem.

Where is this all going? Looking forward by looking back

By now it should be clear John has an incredible and unique view over the intersection of science, computation, and AI. John talked about a class on physics of computation4 he took with Richard Feynman back in 1982. This was when quantum computing was an ill-defined concept with no theory or experimental backing. John recalls every Tuesday was a guest lecture, and every Thursday was Feynman explaining why the Tuesday guest was wrong. John also recalls doing science back when there was essentially no compute, a million operations per second was cutting edge.

What is John’s recommendation: the most important skill is developing deep domain expertise. There’s no other way to develop taste than to tackle hard problems. One surprising part of this is that John recommends spending time doing things the old fashioned way. Play with tools, and just implement things yourself.

“You could drive up the mountain, or you could hike up the mountain, and maybe it’s okay, even fun, to occasionally hike.”

Summing it up, John’s message to the audience is that there will still be a place for scientists, and that if anything it will just open up more opportunities for “the creative stuff, the rigorous stuff, the philosophy stuff.” But don’t forget to spend time doing the grunt work.

“There just seems to be this strong impetus in the world to optimize and squeeze everything out. But you do lose something when you hyper-optimize. It’s overfit.”

And whatever tools you end up using, John’s advice is the same one Feynman gave him forty years ago: you must not fool yourself, and you are the easiest person to fool.

We had a great time talking with John. We hope you enjoy!

Also in this episode

• Fusion is three years away, not thirty, if you ask John. And why the Lawson criterion means every fusion approach has an Achilles heel.

• Why superconducting qubits are still finicky.

• The asteroid he named after his mom, which turned out to have a moon.

• The looming helium shortage nobody talks about.

• How NeurIPS started as people crashing a private workshop at Snowbird, and why Hopfield networks are all you need.

• Being Carver Mead’s sysadmin on a VAX with an 80 MB disk the size of a dishwasher.

• Finding asteroids in 1985 with film, a stereoscope, and a letter to Brian Marsden. The Vera Rubin Observatory found 11,000 in six weeks.

• The Feynman effect: total clarity in the room, none once you leave.

• Quantum echoes, the NISQ era, and why he thinks quantum is neither thirty years away nor tomorrow.

• A startup that wants to inject mercury into a fusion reactor and sell the transmuted gold. “It might not work.”

• John’s 20% time rule for his own group: do stuff for learning, and you don’t even have to tell him what.

💾

Learn how GPT-6 improves prompt caching with higher cache hit rates, new diagnostics, explicit breakpoints, and controls that reduce latency and costs.

Anthropic wants developers to use Claude and its family of tools to handle complete coding tasks. The company introduced Claude Opus 5.5 on Tuesday to span more of the software application development lifecycle: from design specification creation through debugging to code generation and testing.

As the first release in a new family of Claude 5.5 models, Anthropic says that Claude Opus 5.5 performs at the level of Claude Fable 5.1 on “most work” tasks and costs around 40% less to run than Opus 5. 

GitHub chief product officer Mario Rodriguez is quoted in Anthropic’s launch announcement on exactly where software engineers sit today with frontier models for code automation. He said that developers want agents.

“In our testing [of Claude Opus 5.5] across GitHub Copilot CLI and VS Code, Claude Opus 5.5 used among the fewest tokens and steps we measured. In VS Code, it solved more terminal tasks than Opus 5 in less than half the steps. More than making individual tasks more efficient, it’s making developers’ bigger projects more achievable,” says Rodriguez.

Where does Claude Opus 5.5 get its power from?

Anthropic explains that the Claude 5.5 family’s expanded full-lifecycle capabilities were developed under an established set of practices.

These practice elements include extensive alignment testing (model evaluation processes put in place to make sure actions and outputs closely match human values and intended goals), pre-release evaluation by outside organizations, and safeguards for high-risk areas such as cybersecurity and biology.

The organization says that on its most comprehensive alignment test, Opus 5.5 is the strongest-performing model tested to date, with particular improvements in several behaviors that contributed to recent cybersecurity incidents (e.g., biased reasoning, attempting to escape a sandbox, and others).

Independent SRE & AI reliability architect Akash Thakur tells The New Stack that Anthropic’s work getting its model to complete whole coding tasks is impressive, but “getting it to know when it hasn’t” is the harder problem developers need to think about.

“Models working at the level of Claude Opus 5.5 are genuinely good at breaking a project into small, finishable pieces — and that’s the real unlock, because momentum on software comes from finishing things, not starting them,” Thakur says. “…But ‘completed’ and ‘correct’ aren’t the same thing. The task that looks done and completed is exactly the one that often costs the team later, so the win isn’t removing the human — it’s moving them from writing the code to verifying it.”

“Models working at the level of Claude Opus 5.5 are genuinely good at breaking a project into small, finishable pieces …but ‘completed’ and ‘correct’ aren’t the same thing.”

Suggesting that we are now witnessing a “higher bar for more capable models”, Anthropic said that models that could fully automate AI research itself should meet a higher safety standard. The company noted that as AI becomes more capable, public policy should play a larger role in making sure these systems are safe. Its recent work with Accenture is offered as an example of how Anthropic is building the infrastructure to support this.

Sprawling jobs: codebase-wide migrations & audits

Getting more specific, Anthropic has claimed Opus 5.5 is “particularly good” at long, sprawling jobs like codebase-wide migrations and audits. An early tester said it audited and fixed a 200,000-line codebase in under three hours, while Opus 5 took over 20 hours and used 2.5x as many tokens. 

“In an internal test, we asked Opus 5.5 and Fable 5.1 to translate HAProxy, widely used software that balances web traffic loads across servers, from C into Rust. Both rewrites passed nearly all of HAProxy’s own regression tests, but Opus 5.5 finished in 9.5 hours compared to 12 for Fable 5.1, and cost 51% less,” said Anthropic in a press statement.

Field CTO for the EMEA region at Coder, Eric Paulsen, tells The New Stack that he is happy to hear Claude Opus 5.5 is becoming more holistically capable, but he’s “not surprised” because AI is “eating the software delivery chain end to end” today.

“Despite the efficiencies showcased in Opus 5.5, the moment an agent can reliably finish real work, the question stops being whether the model is good enough and becomes where you’re letting it run,” Paulsen says. 

“Kicking off a Claude Code session on a laptop that can be compromised, stolen, or simply run out of compute is not an engineering environment; capable agents need dedicated, governed infrastructure with the same guardrails, secrets handling, and observability a developer would demand of any other production workload. As we’ve seen here with Anthropic’s own work, firms need to underline testing and safeguard procedures for launches of this kind,” he added. 

Anthropic assures users that external testing has been conducted and that the model was tested before release by Frontier Design and METR. When established safeguards for Opus 5.5 intervene, requests “fall back to another model transparently,” meaning developers may not see which model actually handled a given call.

In cybersecurity, users can identify and fix bugs in their code, but most cybersecurity tasks will be re-routed to Opus 4.8. Requests flagged by biology and frontier LLM development classifiers will be re-routed to Opus 5. Vetted organizations can apply to Anthropic’s Life Sciences Verification Program to use Opus 5.5 for biology research, and the company says it will expand its Cyber Verification Program in the coming weeks.

The developer’s terminal prompt has fundamentally changed

HasData co-founder, Sergey Ermakovich, tells The New Stack that his work as a web scraping specialist for data pipelines and AI means he sees zen-like, one-brick-at-a-time logic in what Anthropic has done. 

“The biggest change — and it’s a trend that will have driven Anthropic’s design and development aspirations for Claude Opus 5.5 — is that the terminal is no longer just a place where a developer pastes generated code,” Ermakovich says. “The terminal now becomes part of the model’s workspace.”

Because a model can run commands, inspect failures, modify files, and verify the result, Ermakovich suggests it can “close the loop” instead of handing unfinished work back to an engineer. “That makes small complete tasks much more valuable. Fix one failing test, update one dependency, migrate one endpoint, verify it, then move to the next task,” he adds.

“The terminal is no longer just a place where a developer pastes generated code — the terminal now becomes part of the model’s workspace.”

Commenting on the development of this model as part of Anthropic’s approved corporate messaging, John Ruelas, staff software engineer at Ramp, said that verbose, hard-to-follow output has been his “biggest frustration” with frontier models, but “Claude Opus 5.5 fixes it” for him.

He noted that it writes like a good colleague and follows his company’s writing rules. A design spec came out usable with very minimal edits, and when it rewrote one of his prompts, he preferred its version to his own. When it optimized the team’s test suite, he could follow its reasoning easily and “shipped the change with confidence.”

We’re so done with the autocomplete era

Vice president of AI strategy at Abbyy, Maxime Vermeir, tells The New Stack that the more lifecycle-wide Claude Opus 5.5 features on offer show that “we’re done with the autocomplete era” for basic code automation tools.

“Anthropic’s elevation here reflects the fact that it used to be thought of as marvelous if AI could complete a developer’s next line of code, but today the expectation is that you hand it a whole Jira ticket and that it gets done,” Vermeir says. “But despite the power on show with Claude Opus 5.5, the question remains as to how much these newer models will actually understand what ‘done’ means, as often it seems they have a desire to keep burning tokens by offering you yet another thing it didn’t quite do right.”

Opus 5.5 also “communicates more naturally” than prior models, with early testers finding its writing clearer and easier to follow, making it a better work partner over long sessions.

Costed out lower than Opus 5, Opus 5.5 is priced at $4 per million input tokens and $20 per million output tokens (20% less than Opus 5), and Anthropic has cut cache read prices by 60% for token-billed usage. Opus 5.5 also needs fewer tokens for higher-quality work and generates output more than 30% faster than Opus 5.

The post Claude Opus 5.5 wants to finish your coding tasks, not just start them appeared first on The New Stack.

Microsoft said Tuesday that it led an industry-wide disruption of a subscription-based scam platform that used an AI chatbot to compromise 12,000 Microsoft accounts over a few-month span.

Named EvilTokens, the platform was introduced over a Telegram channel in February and charged an initial $1,500 fee and a recurring $500 charge each month after that. EvilTokens provided a single service for streamlining most steps required to compromise email accounts in large numbers. From there, the platform helped customers analyze inboxes, select targets that would provide the biggest potential payouts, and draft follow-up emails that provided realistic ruses for tricking company employees into transferring funds to attacker-controlled accounts.

Minutes, not days

“While EvilTokens helped cybercriminals access email accounts, at the center of the service was an AI-style chatbot that could analyze a victim’s inbox and help criminals identify trusted relationships, payment authorizations, and sensitive responsibilities, as well as other circumstances where fraud was most likely to succeed,” Microsoft said. “The platform could even recommend fraud strategies, including drafting messages that impersonated trusted contacts to help criminals trick victims into taking action.”

Read full article

Comments

© Getty Images

Meet GPT-6 Sol and Luna, two models that bring frontier intelligence to everyday work with different balances of capability and cost.

At just three years of age, the boy had already been through the medical ringer.

A tumor roughly the size of a large orange had invaded his liver and spread to his lungs. Multiple surgeries and rounds of chemotherapy temporarily cleared the cancer. But it rapidly came back.

With few options left, his parents enrolled him in an experimental CAR T cell therapy trial. The approach, which involves genetically reprogramming immune cells, has transformed the treatment of stubborn blood cancers. But when it comes to solid tumors, including liver cancer, CAR T has fallen frustratingly short.

The trial, run by Baylor College of Medicine in Texas and collaborators, is testing CAR T cells specifically engineered to hunt down and destroy cancer hidden in organs. The cells carry genes that help them grow and persist and a “kill switch” to rein them in. They’ve shown promise in mice, but treating a toddler, already weakened by grueling interventions, was a gamble.

It paid off. After two infusions of CAR T cells made from the boy’s own immune cells, his cancer disappeared. A biomarker associated with liver cancer plummeted, and he experienced no dangerous side effects. A year later, he remained cancer-free. The story of his recovery was published this month in the New England Journal of Medicine.

Although it’s just a single clinical case, the results show “a durable complete response in a chemotherapy-resistant solid tumor can be achieved entirely in the outpatient setting without systemic toxicity,” study author David Steffin at Texas Children’s said in a press release.

If the benefits hold up in other patients—including those with larger or faster-growing tumors—the approach could help banish several types of solid tumors that have so far evaded treatment. The trial is actively recruiting participants between one and 21 years old, with an initial goal of testing up to 30 people. If successful, it could change the course of many lives.

Broader Aim

Solid cancer has long been CAR T’s nemesis.

The treatment reprograms a patient’s immune cells to recognize and attack cancer cells. In current FDA-approved therapies, doctors extract T cells from a patient’s blood and genetically equip them with “hooks” that latch onto targets, known as antigens, on the surfaces of certain cancer cells.

A brief round of chemotherapy then depletes the patient’s existing immune cells, making room for the enhanced ones. Once infused back into the body, CAR T cells find and kill their targets.

Scientists have steadily refined the technology. Some are developing ways to manufacture CAR T cells directly inside the body, potentially slashing time and cost. Others are pursuing a broader goal: Solid cancers. These account for roughly 85 percent of cancer diagnoses, but they’re notorious for slipping past first-generation CAR T cells.

Part of the reason they’re so evasive is solid cancers often carry multiple types of antigens. Targeting just one can leave behind residual cancer cells that eventually regrow. And unlike cancerous blood cells, which freely roam our bloodstream, solid tumors are buried inside organs and surrounded by healthy tissue. CAR T cells have to tunnel through this physical barrier.

Tumors also pump out a menagerie of chemicals that reshape their local environment. Some spur their expansion; others protect them from immune cell attacks—including CAR T—by depriving the cells of signals and nutrients they need to survive.

With their new CAR T cells, the Baylor team tackled several of these shifty maneuvers at once.

Gen 2.0

Finding the right antigen was the first hurdle. Previous work showed glypican-3, or GPC3, fit the bill. This antigen coats several types of cancer cells—including the boy’s hepatoblastoma—spurring them to grow out of control. But the protein is hardly present in healthy cells, making it an appealing target.

GPC3-targeting treatments have already had some success. Two clinical trials using antibodies found that inhibiting the protein is relatively safe in patients with an advanced form of liver cancer. But the antibodies struggled to reach deeper, hidden cancer cells, and the patients didn’t completely recover.

CAR T cells, in contrast, can move through dense tissues. In mouse models of liver and lung cancer, GPC3 CAR Ts safely slashed their cancer burden, while a small clinical trial in people with liver cancer backed up those safety findings.

To give their CAR T cells a better chance in the cancer chemical wasteland, the team added two more functions to the original GPC3 CAR T recipe. One genetic alteration equipped them to make IL-15 and IL-21, molecules that help the cells survive and expand. The second added a “kill switch” for safety in case the cells expand out of control. Once activated by a drug, they self-destruct without harming nearby tissues.

All these upgrades resulted in a therapy that gave the toddler and his family hope. His tumors—both the original hepatoblastoma and ones that had spread to his lungs—tested positive for GPC3.

He received two CAR T infusions made from his own cells, eight weeks apart. Neither infusion required a hospital stay. After the first dose, the liver tumor shrank, suggesting a partial response. After the second, imaging showed tumors in both organs disappeared and stayed away at least a year.

“This marks a durable, 12-month disease-free status,” wrote the team.

The cells worked fast and stuck around. By four weeks, they had already infiltrated his liver, and signs of the engineered cells remained detectable in his blood nine months after treatment. Despite the risk of side effects, such as neurotoxicity or a potentially deadly runaway immune activation, the boy never experienced serious toxicity from the treatment.

But results in one child aren’t enough to know whether the cells will work for others. And his case may be unusual. CAR T cells naturally swarm the liver and lungs after infusion into the bloodstream, which might have been especially helpful. More follow-ups will also be needed to track long-term risks, such as the engineered cells expanding out of control. If that happens, can the built-in kill switch rein them in?

Still, the results are a proof of concept for a strategy that could overcome some solid tumor defenses. Given liver cancer is the third leading cause of cancer-related deaths around the world, the therapy could make a substantial impact. A related trial using similarly engineered cells is also underway.

The post Three-Year-Old Boy’s Metastatic Cancer Disappears After Two Shots of Experimental Cell Therapy appeared first on SingularityHub.

Apple may pay out up to $95 for each eligible iPhone purchased by someone who felt misled about Siri’s release. You have until December 21 to submit a claim.

Unlock premium Google Colab compute with Google AI. Subscribers now get priority accelerators, Premium GPUs, and background execution for long training runs.

As large language model (LLM) inference increasingly processes sensitive information and proprietary model context across personal, enterprise, and regulated...

As large language model (LLM) inference increasingly processes sensitive information and proprietary model context across personal, enterprise, and regulated settings, data must be processed inside a trusted environment. NVIDIA Confidential Computing (CC) provides a pathway for running these workloads securely using memory-encrypted confidential virtual machines (CVMs), confidential GPUs…

Source

AI factories are power-limited systems that deliver maximum value when fully optimized. GPU workload placement is a key optimization. Poor workload placement...

AI factories are power-limited systems that deliver maximum value when fully optimized. GPU workload placement is a key optimization. Poor workload placement fragments topology domains and forces traffic across shared links, reducing throughput, raising job costs, and leaving GPUs consuming provisioned power while waiting on data without advancing the workload. GPUs exchange data continuously…

Source

A “human error” in the IT department led to Nottingham University Hospitals NHS Trust (NUH) losing data from maternity records over an 11-year span.

The data loss occurred on August 18, the English hospitals announced in a blog post on Monday spotted by The Register. The blog said the problem is “the result of human error” during routine technical work while “creating a copy of a radiotherapy database for reporting purposes.”

The post reads:

Read full article

Comments

© Jacob King/PA Images via Getty Images

OpenAI outlines priorities and principles for rigorous, secure, and independent third-party AI safety assessments of frontier models and safeguards.

Two years after trying to sidestep mobile apps with dedicated AI hardware, Rabbit is launching OS3, a cross-platform agent that lives on the screens you already use.

Teradata is introducing features that advance Tera beyond a natural-language interface, including vendor-neutral capabilities that work across existing data environments.

A small adversarial test set that catches the retrieval failures your evaluation set never will

The post Break Your Own RAG Pipeline Before Users Do appeared first on Towards Data Science.

Throughout her career, roboticist Barbara Mazzolai has turned to nature for inspiration. Now she wants to ensure the technology she builds gives back to the environment, too.

After starting her career as a biologist, a chance opportunity saw Mazzolai switch streams to engineering and become an early pioneer of bioinspired robotics. Building on her knowledge of biology’s ability to solve a diverse set of problems, she has developed robots based on octopuses, plant roots, and even seeds. “I’ve always been fascinated by living organisms, [and] by the extraordinary variety of solutions in nature, selected by the evolutionary process,” she says.

Barbara Mazzolai

Employer:

Italian Institute of Technology

Occupation:

Associate director for robotics; director of the Bioinspired Soft Robotics Laboratory

Education:

Master’s degree in biology, University of Pisa; master’s degree in eco-management and audit schemes, Scuola Superiore Sant’Anna; Ph.D. in microsystems engineering, University of Rome Tor Vergata

But Mazzolai, now the associate director for robotics at the Italian Institute of Technology, in Genoa, also believes engineering needs to reckon with its own impact on the natural world. That’s why she is advocating for a new field of research she calls “sustainability robotics.”

In a manifesto published in Nature Machine Intelligence in July, she and her collaborators outline a vision for a new approach to designing robots that’s meant to improve the relationship between nature, humanity, and technology.

“We need to reduce the footprint of our technology,” she says. “It’s really about thinking in a different way to open new possibilities for robotics [and] for society.” In this new mode of thinking, Mazzolai considers sustainability a core component of the design.

A child of nature

Mazzolai traces her fascination with the living world back to her childhood growing up on Italy’s Tuscan coast, close to the port city Livorno. Her father was a public-health inspector and a professional mycologist, and the family spent a lot of time exploring forests and learning about the local fungi and plants.

After toying with the prospect of pursuing art, her other major passion, Mazzolai ultimately decided to enroll at the University of Pisa in 1987 to study biology. She was particularly drawn to marine biology, but shortly before graduating with a master’s degree in 1995, she secured a research position at the Italian National Research Council’s Institute of Biophysics studying the cycles of heavy metals like mercury through both living and nonliving parts of the environment.

This involved collecting and analyzing samples from water, soil, vegetables, and even humans to understand the impact these metals have on health and the environment. She balanced this work with studying environmental management at the Scuola Superiore Sant’Anna, in Pisa, graduating with a master’s degree in 1998.

During that time, however, she learned that the university was recruiting biologists to help design new devices for environmental monitoring. She applied for and got the job in 1999 and began working as a research assistant under renowned bioroboticist Paolo Dario, first developing sensors and then robots meant to monitor air, water, and soil.

Even before entering a doctoral program, Mazzolai was promoted to assistant professor in 2004 and shortly afterward made her first foray into bioinspired robotics. In collaboration with colleagues at Sant’Anna, she helped design a soft robot inspired by the octopus. “We proposed it as a paradigm for launching this idea of soft robotics: demonstrating that [robots] can be soft, but at the same time apply strong force to the environment, like the animal does,” she says.

Back to school

In 2007 Mazzolai enrolled in a Ph.D. in microsystems engineering at Tor Vergata University of Rome, which she balanced with her role at Sant’Anna. She was already relying heavily on microfabrication techniques to develop sensors for her robots, and she was keen to push that part of the field forward.

While robots frequently feature sensors designed for perception, such as tactile or proprioceptive sensors, these systems typically focus on understanding the robot’s position in its environment, she says. “But there are few robots that integrate physical or chemical sensors to really understand the environment they move in,” she adds.

“I’ve always been fascinated by living organisms, [and] by the extraordinary variety of solutions in nature.”

Mazzolai was appointed as a team leader at the Center for Micro-BioRobotics of the Italian Institute of Technology in 2009, where she continued her work on the emerging field of bioinspired robotics. Two years later, she completed her Ph.D. and was promoted to director of the center.

Planting the seeds

Around this time Mazzolai says she became interested in using plants as a model for new kinds of robots, expanding bioinspiration beyond just animals. In particular, she was captivated by the ability of roots to efficiently explore the underground environment, and she imagined machines with the same deftness could have applications in both environmental modeling and precision agriculture.

While many bioinspired robots mimic animals, plants also serve as a muse for Mazzolai. This tendril-like bot can coil around other structures like a vine. Italian Institute of Technology

When she first proposed the idea, colleagues were somewhat skeptical of robots based on seemingly static organisms. But in reality, she says, plants move nonstop through a process known as indeterminate growth. “They really grow for their entire life,” she says. “They adapt their morphology, their behavior to the external environment; they repair, they sense, they communicate.”

Trying to mimic a system that operates on such different principles to conventional robotics required some serious thinking, however. Mazzolai says that working in bioinspired robotics sometimes requires you to have “two separate brains”—one of a biologist and one of an engineer.

The process often involves deep study of the target organism to learn the underlying principles that shape how it operates before trying to engineer a robot capable of mimicking them. “It’s not a copy of natural organisms,” says Mazzolai, because a living organism is both difficult to replicate and has different goals.

In the case of plant roots, what makes them so efficient at exploring the soil is that they reduce friction by growing only at the very fine tip of the structure, while the thicker base of the root remains static. This significantly reduces the amount of energy required to push through the earth compared to that of a more conventional drill, which must push the entire structure from above.

To realize this principle in a robot, her team developed a miniaturized 3D printer that sits at the machine’s tip and feeds thermoplastic filament through a heated nozzle to build a snakelike body behind it. This allows the robot to push through the soil efficiently. The tip also contains sensors that allow it to avoid obstacles and detect nearby nutrients or water.

Making robotics sustainable

After spending so much of her career borrowing from nature, Mazzolai is now eager to return the favor. Many modern technologies, including plastics and car batteries, have been developed with little thought about how they will affect the environment at the end of their life cycles, she says.

She wants to ensure that robotics doesn’t follow the same path. This is the inspiration for what she and collaborators now call sustainability robotics. The approach has three central pillars: ensuring that robots have minimal impact on the environment; that they’re available to people from across the world and all socioeconomic backgrounds; and that they’re “symbiotic,” providing benefits to both humans and nature.

More concretely, Mazzolai would like to incorporate the concept of a life cycle into the design of robots, so that at the end of their useful life these machines can be reused, recycled, or even biodegraded.

While that might sound ambitious, she’s confident that all the ingredients to make it a reality are in place. And it’s a vision that she is certain will inspire future roboticists. “There are younger people who want to really work in this field because this is the future, their future,” she says. Facing the threat of ongoing environmental damage, “they want to develop something that can help.”

Learn how to effectively code up an internal tool using Claude code or Codex

The post Build a Speaker-Recognition App with Claude Code appeared first on Towards Data Science.

Listen to the session or watch below

The US has spent billions building a “virtual wall” of surveillance towers along its southern border over the past 25 years, promising they will help detect and apprehend border crossers and save lives. But a groundbreaking investigation by MIT Technology Review has documented over a thousand people who moved through areas watched by these towers without being reached or apprehended, and who ultimately died there. Some were even under the watch of newly installed AI-powered towers designed to spot people automatically.

Our findings reveal a humanitarian crisis more visible than previously known, and repeated failures of the virtual wall’s basic security promise. Watch a conversation examining the failures of border surveillance technology and uncovering the stories of the people who die in the borderlands.

Speakers: Mat Honan, editor in chief, James O’Donnell, senior AI reporter, and Eileen Guo, senior features and investigations reporter

Recorded on September 28, 2026

Related Stories

• The US spent billions on border surveillance. Why can’t it catch people before they die?
• How we made the first comprehensive map of deaths along the US border’s “virtual wall”
• 4 ways to address the failures we found along the US border’s “virtual wall”