ai-pocalypse The National Institutes of Health (NIH) has called for so few in anticipation that the Trump administration will implement an AI action plan to boost the government's use of AI on Wednesday.
Last week, guidance issued to researchers at the NIH, part of the U.S. Department of Health and Human Services (HHS), allowed grant applications created with the help of generator AI.
“NIH will not consider applications that are substantially developed by AI or that contain sections that are substantially developed by AI as original ideas for applicants,” the health agency notification explains.
“If AI detection identifies a post-prize, the NIH may refer the issue to the Research Integrity Bureau to determine whether there is research misconduct, while also taking enforcement measures including cost removal, future award withholding, full or partial grants, and possible termination.”
NIH did not respond to requests for comment, but according to the notification it suddenly received an unusually large number of research applications, some of which appear to have been created with the help of AI tools.
“While AI may be a useful tool to reduce the burden of application preparation, prompt submission of numerous research applications from a single principal investigator can unfairly burden the NIH application review process,” the notification states.
According to NIH, few scientists submit an average of six or more applications per year, but AI tools now submit more than 40 individual research applications in one submission round.
Worse, the gap between the institution's ability to review research applications and the speed at which AI-generated submissions arrive appears likely to grow following natural reports suggesting that many grants face layoffs as many grants are looking to appoint aspiring scientists.
Last month, around 500 NIH staff members signed a petition urging NIH and HHS leadership to stand up to science and academic freedom in the face of Trump administration cuts. Scientists sought to recover the scheduled research grants and the recovery of fired staff.
The NIH says it will ferret research applications generated by AI using unspecified technologies, while also stressing that all such applications must comply with grant policies that expect research institutions and teams to propose original fundraising ideas.
Health agencies acknowledge that AI tools may be appropriate for the limited tasks of preparing research applications, despite warning that AI use can lead to plagiarism, invented citations, and other scientific misconduct.
NIH faces the same flood of overwhelming open source projects such as Curl, Python, Open Collective, and academic publishers, journalism, web search, and social media. If human content ratings meet automated content generation, people either cannot or do not want to keep up because of the poor quality of the generated output.
This dynamic once prompted the Claude AI model family to refuse the use of generated AI by those submitting job applications.
“We encourage people to use AI systems during their roles to help them work faster and more effectively, but please don't use AI assistants during the application process,” the company's job board said earlier this year.
“We want to understand your personal interest in humanity without mediating through AI systems, and we also want to assess your non-AS support communication skills.”
But earlier this month, humanity decided on a more nuanced approach, recognizing the promotional issues that are likely inherent in allowing the use of its own products. Job seekers now say, “While it makes sense, we encourage you to use Claude to show us your unique perspective, skills and experience.” ®
