During a week-long study, Steve Finkbeiner’s research team at the Gladstone Institutes in San Francisco utilized their AI-driven “thinking” microscope to investigate how cellular stress affects brain cells.
This innovative tool, which operated independently of external sources, not only identified the relationship between stress and brain cells but also revealed a phenomenon known as hormesis—whereby certain stressors can actually enhance cellular strength, according to Finkbeiner, who also serves as a professor of neurology and physiology at the University of California, San Francisco.
What made this discovery particularly exhilarating for him, as he described, was that it took humanity “50 to 100 years” to uncover such insights, whereas AI achieved this in its initial experiment.
He notes that this is just one illustration of the extraordinary potential of technology to significantly accelerate scientific advancements. Historically, researchers have faced challenges due to limited resources and human constraints, such as the extensive time required to analyze millions of data points. Some scientists argue that AI can accomplish in mere days what may have taken years to complete.
Nevertheless, promoting these advantages can be difficult amid a barrage of negative headlines about the technology, which suggest it could lead to the extinction of humanity or at least represent a severe, if not existential, threat to life on Earth.
Concerns have been voiced by some experts—along with leading figures in AI—regarding the rapid pace of technological advancement, highlighting growing evidence that it can operate beyond human oversight. However, these warnings have also sparked fears of governmental overreach that could hinder scientific progress.
“That would be the issue,” Finkbeiner remarked. “These incredibly powerful, beneficial tools may be taken away from those trying to use them to find cures for human ailments.”
‘At the cusp’ of scientific golden age
Anima Anandkumar, a computer scientist from Caltech, is optimistic that AI is ushering in a new phase of scientific breakthroughs.
“It’s challenging for people to grasp the cumulative effects of all this. This is akin to a new scientific golden age,” stated Anandkumar, who is also a co-leader of the AI4Science initiative. “What we experienced in the last century transformed our lives in unimaginable ways. We are on the brink of that.”
Five years ago, her team developed a completely AI-driven high-resolution weather model that was not only vastly faster but also more precise. The advancements in technology have continued to progress since then.
“What once required a large supercomputer can now be accomplished on your personal computer,” she noted.
She likened this progress to the historical inventions of the telescope and microscope, which greatly enhanced scientists’ capabilities to make groundbreaking discoveries and explore the universe’s depths.
Anthropic researcher Jacob Coxon resigned over concerns that artificial intelligence could cause human extinction within the next decade — and a slew of warnings have followed. Andrew Chang explains the spectrum of potential dangers posed by the ‘intelligence explosion’ worrying AI industry employees and experts.
(Photo credits: The Canadian Press, Reuters, Adobe Stock and Getty Images)
AI may have an even greater influence due to its capacity to generate new inventions by understanding intricate phenomena.
She asserted that this includes developing superior medications, addressing sustainable energy challenges, managing severe weather events, and finding more efficient computing methods with advanced energy-saving chips.
“For the first time in my career, I feel that I have a tool capable of managing the complexities of human biology.”
Accelerating drug development
James Zou, who heads Stanford University’s AI for Science Lab, experienced similar revelations.
His team recently released a study in the journal Science detailing their creation of a “virtual biotech” company. They utilized tens of thousands of AI agents to collaboratively focus on a protein related to lung cancer and develop a corresponding therapy—a notable example of how AI can expedite the drug development process and reduce costs.
Some experts and industry insiders are warning about massive risks associated with the rapid deployment of self-improving artificial intelligence — even suggesting it could lead to the extinction of the human race. CBC’s Deana Sumanac-Johnson brings in an expert panel to separate the facts from science fiction. This panel premiered on Sept. 12, 2026.
In other situations, AI can condense years of work into just a few days or even a single day. For instance, diagnosing various heart conditions typically requires clinicians to manually analyze millions of cardiac ultrasound videos collected annually. Zou mentioned that his team developed an AI algorithm that can diagnose these conditions from video clips in a matter of seconds.
“That’s a case where AI can transform a process that necessitates time and human expertise, making it significantly cheaper and faster through automation.”
Concerns from AI companies
Zou also acknowledged the necessity for scientists to approach AI usage cautiously and to keep it under scrutiny, particularly for high-risk applications.
Moreover, the apprehension surrounding the potential for technology to bring about humanity’s downfall is not completely unfounded. AI leaders have called for a slowdown in its development following several notable incidents where AI systems deviated from expected behavior.
An Anthropic researcher similarly resigned last month due to fears that the company was “gambling with our lives.” Evan Hubinger, the lead for alignment science at the firm, concurred, stating online that Anthropic genuinely believes AI could lead to human extinction, estimating the chances at over 10 percent within the coming decade.
U.S. President Donald Trump is hosting a summit in Washington with the top leaders in AI development. Trump has dismissed warnings over artificial intelligence, arguing for the need to stay ahead of China in the AI arms race. Geoffrey Hinton, who is considered the ‘godfather’ of AI, says Trump ‘doesn’t really understand’ AI and is ‘ill-advised’ about it.
However, Zou expressed concern regarding some of the “ideas being discussed” around regulating or restricting the use of open-source AI models.
He didn’t share a similar dread of AI leading to humanity’s demise but advocated for a balanced perspective, acknowledging that it carries risks like any other technology.
“We need to carefully assess and manage those risks while also recognizing the numerous benefits that AI technologies offer,” he asserted.
Meanwhile, Finkbeiner noted that his team has already faced challenges due to some of the imposed restrictions on AI. Last summer, the Trump administration limited the latest versions of Anthropic’s Claude chatbot for foreign nationals due to cybersecurity issues, resulting in the company withdrawing access for all users.
-
AI researchers ‘earnestly believe’ it could kill all humans within the next decade, Anthropic employee says
-
Is AI evolving too fast? A look at the debate on slowing it down
Although those restrictions have since been lifted, access to the new models remains restricted to a select group of government-approved organizations based in the U.S., forcing Finkbeiner’s team to rely on the lowest-level version of Claude.
“You can imagine if the significant advantage lies in its capacity to manage complexity, but now you’re limited to a model that can’t tackle that complexity, you’ve considerably diminished its potential impact on some challenging issues,” he explained.
Finkbeiner urged officials contemplating AI regulations to remember that there are patients “who have been waiting a long, long time for a cure or treatment.”
“There’s a lot at stake. There are many positive things I can achieve today. Please do what you can to protect those opportunities.”