How Teachers Can Defeat AI
As someone also in the academic trenches, I appreciated John J. Goyette's op-ed 'How to Stop Students From Cheating With AI' (May 20). My plan for the autumn semester is to offer an old-school approach: Students who take handwritten notes in physical notebooks will be allowed to take exams 'open book.' Hopefully this will disincentivize using computers in the classroom and reinstill in students the value of note-taking. Perhaps this will even lead to the return of some creative doodling.
Prof. Eric Zolov
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Medscape
26 minutes ago
- Medscape
Catching Resistance Early: Can New Breast Cancer Drug Help?
CHICAGO — Can spotting an emerging ESR1 mutation early and changing first-line drugs before progression improve outcomes in patients with hormone receptor (HR)–positive, human epidermal growth factor receptor 2–negative advanced breast cancer? Interim findings from the SERENA-6 trial suggest that may be the case. Patients who switched from a first-line aromatase inhibitor to camizestrant, an investigational next-generation oral selective estrogen-receptor degrader, at the first signs of an emerging ESR1 mutation demonstrated significantly improved progression-free survival compared with those who continued their initial regimen. Notably, circulating tumor DNA (ctDNA) testing allowed investigators to identify ESR1 mutations, which emerge at the time of disease progression in about 40% of patients on a first-line aromatase inhibitor and lead to treatment resistance. Camizestrant, which has shown activity in patients who develop ESR1 mutations, helped improve first-line outcomes and has 'potential to become a new treatment strategy,' according to co-principal investigator Nicholas Turner, MD, PhD, professor and honorary consultant in medical oncology at the Institute of Cancer Research and Royal Marsden Hospital, London, England, who presented the findings at the American Society of Clinical Oncology (ASCO) 2025 annual meeting. Results were simultaneously published in The New England Journal of Medicine . This trial also demonstrated 'the clinical utility of ctDNA monitoring to detect and treat emerging resistance in breast cancer,' said Turner. While praising the findings, others were not convinced that the SERENA-6 results warrant a change in practice yet. 'Based on first-line progression-free survival alone, this could represent a new regulatory approval path,' said invited discussant Angela DeMichele, MD, of the University of Pennsylvania, Philadelphia. But, DeMichele cautioned, 'I cannot recommend the SERENA-6 strategy at this time.' One key reason, DeMichele noted, is that it's too early to tell whether this strategy improves overall survival. If camizestrant is approved based on progression-free survival and quality of life, DeMichele wondered, is it worth going through the ctDNA testing process if the drug doesn't help patients live longer? Paolo Tarantino, MD, a breast oncologist at Dana-Farber Cancer Institute and Harvard Medical School in Boston, echoed this sentiment in a tweet on X: 'Outstanding results, though not ready for clinical practice (yet),' adding that it will also be 'important to take into account financial, psychological, and systemic costs of the strategy.' Using ctDNA to Track Resistance In the study, 3256 patients who had received at least 6 months of treatment with aromatase inhibitors and CDK4/6 therapy (palbociclib, ribociclib, or abemaciclib) received ctDNA testing with Guardant360 CDx every 2-3 months at the time of routine staging exams. Overall, 315 patients who had an ESR1 mutation detected and had no radiologic evidence of disease progression were randomly assigned to either switch from the aromatase inhibitor to 75 mg of camizestrant daily (n = 157) or continue their aromatase inhibitor/CDK4/6 regimen (n = 158). (An additional 233 patients who had an ESR1 mutation detected were not included for a variety of reasons, including disease progression and consent withdrawal.) At the planned interim analysis, the median progression-free survival was 16.0 months in the camizestrant group and 9.2 months in the aromatase inhibitor group (adjusted hazard ratio [aHR], 0.44; P < .00001). At 24 months, only 5.4% of patients who had continued their initial first-line treatment had not progressed compared with 30% of patients on camizestrant. The progression-free survival findings were consistent across clinically relevant patient subgroups. Patients who switched to camizestrant also showed improved time to deterioration in global health status and quality of life — a median of 23.0 months vs 6.4 months in the aromatase group (aHR, 0.53). At the time of the interim analysis, overall survival data were immature, with 20 deaths in the camizestrant group and 19 in the aromatase inhibitor group (HR, 0.91; 95% CI, 0.48-1.73). As for time to second progression, there were 38 events in the camizestrant group and 47 events in the aromatase group, but the findings were also immature. As for adverse events, 60% of patients in the camizestrant group had a grade 3 or higher event, 10% of which were deemed serious compared with 46% in the aromatase group, 12% of which were serious. Neutropenia (45% vs 34%, respectively) and anemia (5% in both groups) were the most common grade 3 or higher adverse events. Only 1% of patients on camizestrant discontinued treatment due to adverse events. Overall, Turner concluded that 'for people with HR-positive advanced breast cancer, the results of SERENA-6 show that camizestrant plus CDK4/6 inhibitor could be a new treatment option to use at the point of ESR1 mutation detection during treatment with first-line aromatase inhibitor plus CDK4/6 inhibitor — but before the cancer grows.' Despite the promising findings, DeMichele highlighted several key unanswered questions and challenges. Notably, will this strategy lead to longer overall survival and demonstrate clinical utility? Overall survival and time to second progression are currently not known, DeMichele said. The trial did not address whether first-line treatment gains would be lost if camizestrant was given in the second-line setting after anatomic progression. DeMichele also noted the high cost and potential anxiety associated with serial ctDNA testing. Overall, 'the full complement of financial, psychological, and systemic costs is needed to fully assess utility and feasibility for implementation,' she added. SERENA-6 was supported by AstraZeneca. Turner disclosed consulting or advisory roles with AstraZeneca, Exact Sciences, Gilead Sciences, GlaxoSmithKline, Guardant Health, Inivata Lilly, Merck Sharp & Dohme, Novartis, Pfizer, Relay Therapeutics, Repare Therapeutics, and Roche. DeMichele disclosed a consulting or advisory role with Pfizer.
New York Times
32 minutes ago
- New York Times
Anthropic C.E.O.: Don't Let A.I. Companies off the Hook
Picture this: You give a bot notice that you'll shut it down soon, and replace it with a different artificial intelligence system. In the past, you gave it access to your emails. In some of them, you alluded to the fact that you've been having an affair. The bot threatens you, telling you that if the shutdown plans aren't changed, it will forward the emails to your wife. This scenario isn't fiction. Anthropic's latest A.I. model demonstrated just a few weeks ago that it was capable of this kind of behavior. Despite some misleading headlines, the model didn't do this in the real world. Its behavior was part of an evaluation where we deliberately put it in an extreme experimental situation to observe its responses and get early warnings about the risks, much like an airplane manufacturer might test a plane's performance in a wind tunnel. We're not alone in discovering these risks. A recent experimental stress-test of OpenAI's o3 model found that it at times wrote special code to stop itself from being shut down. Google has said that a recent version of its Gemini model is approaching a point where it could help people carry out cyberattacks. And some tests even show that A.I. models are becoming increasingly proficient at the key skills needed to produce biological and other weapons. None of this diminishes the vast promise of A.I. I've written at length about how it could transform science, medicine, energy, defense and much more. It's already increasing productivity in surprising and exciting ways. It has helped, for example, a pharmaceutical company draft clinical study reports in minutes instead of weeks and has helped patients (including members of my own family) diagnose medical issues that could otherwise have been missed. It could accelerate economic growth to an extent not seen for a century, improving everyone's quality of life. This amazing potential inspires me, our researchers and the businesses we work with every day. But to fully realize A.I.'s benefits, we need to find and fix the dangers before they find us. Every time we release a new A.I. system, Anthropic measures and mitigates its risks. We share our models with external research organizations for testing, and we don't release models until we are confident they are safe. We put in place sophisticated defenses against the most serious risks, such as biological weapons. We research not just the models themselves, but also their future effects on the labor market and employment. To show our work in these areas, we publish detailed model evaluations and reports. Want all of The Times? Subscribe.
Medscape
37 minutes ago
- Medscape
Biomarker sNfL Falls Short for MS Diagnosis, Experts Say
PHOENIX — Despite its growing role in monitoring treatment response and disease activity in multiple sclerosis (MS), serum neurofilament light chain (sNfL) lacks the specificity needed for diagnosis and will not be included in upcoming revisions to the McDonald criteria, experts say. In addition to monitoring treatment response, sNfL has the potential to eventually play a part in predicting disease course, 'but diagnosis? Not so much,' said Tanuja Chitnis, MD, senior neurologist, Brigham M ultiple S clerosis Center, Harvard Medical School, Boston, at the Consortium of Multiple Sclerosis Centers (CMSC) 2025 Annual Meeting. This not only applies to sNfL but other fluid biomarkers, including glial fibrillary acidic protein (GFAP), which has enormous promise for tracking and understanding the course of MS but also needs context and has important limitations. 'We do not yet know for sure what the updated McDonald criteria will recommend specifically, but sNfL is not very helpful for diagnosis,' said Mark Freedman, MD, director of the Multiple Sclerosis Research Unit at the University of Ottawa, Ottawa, Ontario, Canada. He, like Chitnis, does not see sNfL as a diagnostic tool either as an isolated value or in the context of other currently available signs of disease. Most, if not all, patients with a diagnosis of MS who are eventually confirmed typically have elevated levels of sNfL, but Freedman, who led the symposium on fluid biomarkers said this information is not helpful. And sNfL, he noted climbs with neurodegeneration, but this is not unique to MS. Many other diseases associated with neuronal damage, including amyotrophic lateral sclerosis, are linked to much higher sNfL levels at diagnosis, underscoring its poor specificity, said Freedman, who has led consensus papers on its use. The most recent was published in 2024 on behalf of the CMSC. An Indicator of Response Chitnis said sNfL may eventually play a role in MS diagnosis when combined with other variables, including fluid biomarkers, but for now, its primary value lies in disease management, where it serves as a sensitive tool for monitoring treatment response. A sensitive indicator of disease activity sNfL provides information beyond MRI, said Chitnis, who tracks levels every 3-6 months after initiating or switching therapy. A lack of decline may signal inadequate disease control. Freedman echoed this view, noting that if levels don't drop within a few months of starting treatment, he begins to question its effectiveness — even before other signs of treatment failure appear. 'If I do not see sNfL fall within a few months of starting therapy, I start questioning the therapy,' he said. The clinical value of sNfL was highlighted in a 2023 CSMC consensus paper, which endorsed its use for tracking recovery after relapse, guiding therapy decisions, and clarifying discrepancies between MRI and clinical findings. Since then, evidence supporting its use has only strengthened, Freedman said. 'The tests are now accessible everywhere,' he said. He also noted that there is growing confidence that sNfL is a substitute for MRI in many specific scenarios, such as when access to MRI is limited by cost or distance to an imaging device. Multiple sNfL assays are available, but they are not interchangeable. They vary in sensitivity, turnaround time, methodology, and cutoff values — so any changes in sNfL levels must be tracked using the same assay throughout. 'We have to be careful about using apps to interpret the results,' Chitnis cautioned, noting that such tools are now commercially available. She added that interpretation also depends on the reference population and can be influenced by factors like body mass index, renal function, and age. 'A Novel Window Into the Brain' The value of sNfL in monitoring treatment response — and in tasks such as assessing disease activity in patients with MS who have discontinued DMT — may be just the beginning for this biomarker, which is under active investigation for a range of additional uses, said Chitnis. She noted that the clinical utility of fluid biomarkers like sNfL is likely to expand significantly as researchers learn how to interpret them in combination. In several studies, pairing sNfL with GFAP has offered new insights into progressive MS, particularly when the two markers trend in opposite directions. Despite an imperfect correlation, patients with high GFAP but low sNfL, relative to the other way around, 'probably represent our nonactive secondary progressive MS disease cohort,' Chitnis said. 'This is getting at our understanding of who might be more progressive in regard to the loss of the glial structure in the brain,' she added. Meanwhile, other potential biomarkers associated with increased clinical and radiological MS activity are showing promise. In a recent study led by Chitnis, a panel of 20 such proteins correlated more precisely with gadolinium lesion activity, clinical relapse, and annualized relapse rate than sNfL alone. Ultimately, fluid biomarkers like sNfL are providing 'a novel window into the brain' relevant not just to monitoring disease but potentially revealing the sequence of underlying pathophysiology of MS, said Chitnis. These appear to differentiate relatively benign disease from events leading to the irreversible damage of late stages progression.
