Doctors Were Worse at Spotting Cancer After Leaning on AI, Study Finds
The study looked at four endoscopy centers in Poland, tracking the success rates of detecting colon cancer for three months before AI tools were introduced and three months after. Once AI was introduced, colonoscopies were randomly assigned to either receive AI support or not. The researchers found that doctors who gave colonoscopies without AI after having its assistance available saw their detection rates drop, producing outcomes 20% worse than what they were before AI was introduced.
Making the results all the more troubling is the fact that the 19 doctors who participated in the study were all very experienced and had performed more than 2,000 colonoscopies each. If those doctors can fall prey to de-skilling, seeing their own abilities erode because of reliance on AI tools, the outcomes from inexperienced doctors could be even worse.
There is little doubt that AI tools can help in medical settings. There have been numerous studies that suggest that AI can facilitate everything from the detection of cancers to the diagnosis of illnesses based on a patient's medical history. Analyzing information based on a whole wealth of previous examples is kinda the bread and butter of AI (you know, as opposed to generating braindead content slop), and there is evidence that suggests that humans can augment their own abilities by using AI tools. Studies in medical settings have found that doctors who use these tools can produce better outcomes for their patients.
But no one, including doctors, is immune to the risk of shutting your brain off and relying on AI rather than their own skills. Earlier this year, Microsoft published a study that found knowledge workers who lean on AI stop thinking critically about the work they are doing and feel confident that the assistance from AI will be enough to get the job done. Researchers at MIT similarly found that relying on ChatGPT for essay writing resulted in less critical engagement with the material. In the long term, there's a real risk that reliance on AI will erode our ability to problem solve and reason, which is not ideal when AI continues to generate bad information.
The American Medical Association found that about two in three physicians have already adopted AI to augment their abilities. Hopefully, they're still able to identify when it does something like hallucinating a body part that doesn't exist.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Bloomberg
8 minutes ago
- Bloomberg
Osika: Lovable Margins Are Improving as AI Costs Decline
As the cost of AI declines, European unicorn Lovable's margins are improving, says CEO and co-founder Anton Osika. He adds that while AI will force people to relearn what tools they use, coders using Lovable are at an advantage when using the AI and are seeing increased productivity. Anton Osika, Co-Founder & CEO of Lovable spoke to Bloomberg Daybreak Europe anchor Tom Mackenzie. (Source: Bloomberg)
Yahoo
17 minutes ago
- Yahoo
Bitcoin breaks new record as analysts anticipate surge to $200,000
Bitcoin surged to a new record trading at $124,529 on Coinbase on Wednesday evening, but market watchers are already anticipating another all-time high. Mena Theodorou, co-founder of Australian exchange Coinstash, said that US President Donald Trump's executive order to allow Bitcoin and other cryptocurrencies to be part of people's retirement plans will open up a $12.5 trillion market for digital assets. 'The move toward $150,000 and beyond could arrive faster than many expect,' Theodorou said in comments shared with DL News. By unlocking US 401(k)s, crypto companies will not only be able to offer services to some 90 million accounts, but it also signals that other countries may follow, Theodorou argued. 'Where the US leads, the rest of the world often follows,' he said. Theodorou isn't the only one bullish about the rally driving the price higher. Bernstein, Bitwise and Standard Chartered have all predicted that the price will hit $200,000 before the end of the year. Several factors have driven the rally, apart from Trump's executive order. Steady inflows into spot Bitcoin exchange-traded funds, over 200 companies adding cryptocurrencies to their treasuries, and a weakening US dollar are the key drivers that fuelled Bitcoin's march to it's record high price, cooler-than-expected US inflation, analysts noted. Emir Ibrahim, analyst at crypto liquidity provider Zerocap, said the surge marks 'a shift from the retail-driven hype of past bull runs to a market shaped by institutional adoption, real-world integration, and deep global liquidity.' Another key driver is the growing anticipation that the Federal Reserve will serve up a jumbo rate cut in September. US Treasury Secretary Scott Bessent told Fox Business that the Fed should begin an easing cycle with a 50 basis-point rate cut next month. Lower interest rates support risk-on assets like crypto by making capital cheaper and pushing investors away from bonds and toward higher-yielding alternatives. Bitcoin is not the only cryptocurrency that's benefitting from these drivers. Ethereum is also up this week. It reaching $4,788 on Wednesday, just $100 shy of its all-time high set back in 2021. Crypto market movers Bitcoin is up 1.5% over the past 24 hours to trade at $121,765. Ethereum is up 2.3% to trade at $4,734. What we're reading Trump debanking order will have limited impact on crypto, experts say— DL News Stablecoin Blockchains Are Coming. Here's Why These Two Giants Should Be Nervous — Unchained World Liberty Financial token valued at $0.20 in Trump crypto project's ALT5 treasury deal— DL News Altcoin Interest: Bigger Than 2021 — Milk Road rival Bags hits $1bn in trading volumes — DL News Lance Datskoluo is DL News' Europe-based markets correspondent. Got a tip? Email at lance@ Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
Fast Company
38 minutes ago
- Fast Company
AI is eating change management
Change management is a multibillion-dollar industry built on the fundamental claim that most people dislike change, and that someone needs to manage that resistance. But after decades of organizational theories and billions in consulting fees, the industry does not work as promised: change management projects have a failure rate of around 70%. There's a reason no one asked McKinsey or any other leading consulting firm to run DOGE. As enterprises large and small grapple with the wholesale transformation that will be wrought by the rise of artificial intelligence, it's time to face an uncomfortable truth: Change, especially today, doesn't happen in neat phases. It's cyclical, unpredictable, and requires constant adaptation. The Old Model Never Worked–But It Especially Doesn't Now Traditional change management follows a predictable model. Under investor scrutiny—or to avoid it—the CEO announces a transformation is coming. Consultants conduct surveys, present slides at workshops, and create communication plans to deal with signs of revolt. Success gets measured by stakeholder-focused metrics like 'adoption rates,' and KPIs like 'number of training programs deployed.' It's an industry that prioritizes rationality at the expense of inspiration and serendipity. AI will be the force that kills this episodic approach. First, the tech is already moving too quickly for rigid approaches to be relevant. Second, AI requires significant amounts of training and customization to be effective in most organizations, rendering a 'one-size-fits-most' approach obsolete. And finally, the human side of AI-driven change is more complicated than a standard reorganization—because AI anxiety strikes at the heart of what is human, and what sort of careers we and our children will have. A Tale of Two Companies In our recent work inside companies that are adopting new AI tools and workflows, we've seen the potential for a new way of working. Instead of a traditional change management approach, smart leaders today are understanding—and embracing—that change in the era of AI is often organically driven by shifts brought about by AI eureka moments. Competitive advantage is built not by how quickly you move humans through a change program, but how seamlessly your organization's source code—the unique combination of people, process, and technology—rewrites itself in real time. A Tale of Two Companies Consider a recent tale of two companies. First, fintech company Klarna's recent initiative to automate its customer service operations using generative AI. The company publicly claimed that its AI tools were performing the work of 700 full-time agents, leading to a dramatic reduction in hiring and headcount. The rollout was managed through a centralized, top-down approach: executive-led messaging, internal dashboards to track AI performance, and a focus on cost savings and productivity metrics. But the transition sparked internal unease and external criticism, and Klarna quietly began rehiring human agents within the year. The AI may have delivered efficiency on paper, but the rigid implementation and lack of human-centered change management eroded trust, both inside and outside the company. Contrast that with one of our clients—a multinational pharmaceutical organization that took a radically different approach. Rather than relying on static KPIs and sequential rollouts, they used using AI to surface real-time insights from employee sentiment, social media behavior, and internal feedback loops. These insights continue to inform tailored interventions across roles and geographies. AI-powered chatbots enable employees to access personalized resources on demand, while leaders use behavioral analytics to trigger timely nudges and adapt strategies instantly. The result has been a more agile, inclusive transformation—where change has been continuously shaped by how employees are actually working. How Organizations Can Stay Ahead In this new world, 'best practice' changes from week to week. But the most important trends we see in recent, successful transformations are: First, build a non linear approach. When it comes to generative and agentic AI, you often don't know your best use cases until you experiment. Embrace the 3-D problem solving that comes with transformation by moving to organized but flexible processes that account for two-way feedback. Second, create pilots. Understand that new processes, technologies, and workflows will work differently for each organization and team. Select specific organizational areas for focused experimentation and training. Give them deadlines and establish feedback loops between pilot participants and the transformation team. Then, scale successful approaches across the organization using champions as advocates for the technology and its impact. Third, work to understand and activate teams with precision. Identify specific employee categories to play a role in championing change. Every organization has a group of early adopters—the 'weekend warriors' who explore AI on their own time. And every organization also has laggards—those who will require structured protocols and personalized training plans to implement new systems. Focus your communications—and your expectations—by identifying each group and understanding the different needs it requires. Finally, empower leaders. Measure success not by who attended the meeting or did the training, but by who's actually creating new pathways in process or technology. Encourage those leaders, from the CEO down, to show how they use AI tools, and arm with appropriate nudges for staff. According to the Boston Consulting Group, the small minority of companies already operating at this level are realizing 1.5× revenue growth, 1.6× shareholder returns, and 1.4× ROI. The goal is improving organizational metabolism so your organization stays healthy, instead of contracting a disease that needs treatment. The business model for change management consulting may shift to something far more organic: Enabling leaders to role model and guide, designing teams built for experimentation and imbuing organizational culture with a growth mindset. Adaptability counts most Corporate America rewards risk-taking and stories about explosive growth, rapid innovation, and bottom-line-enhancing layoffs. But it's adaptability that will count most in the AI era, and continuous improvement is what will deliver it. Organizations that continue to rely on traditional change management consultancies are not just wasting money—they're actively handicapping their ability to compete in an increasingly dynamic business environment. Consultants can either change— the irony!— or go down with their ship.
