The US government's new America.gov AI chatbot, powered by Gemini and xAI's Grok, has stopped answering political questions about elections after it initially contradicted President Trump's claims about the 2020 election and inflation. Democratic lawmakers had praised the chatbot for its honest responses, but within a day of its launch, the White House said America.gov will continue providing accurate information while steering clear of election-related queries.
When tested on health questions, the chatbot gave mixed results. A reporter sent 37 health questions and found it sometimes failed to mention court orders restoring CDC recommendations. It also incorrectly advised following a December CDC page about hepatitis B shots for newborns, ignoring a judge's ruling that restored the older schedule. On urgent topics like chest pain, it correctly told users to call 9-1-1.
Key Takeaways
- America.gov AI chatbot powered by Gemini and Grok stopped debunking Trump claims about the 2020 election and inflation
- The chatbot gave mixed health answers, failing to mention court orders restoring CDC vaccine recommendations
- Only 18 of Harvard's 50 concentrations have department-wide policies on AI use in senior theses
- Fulton County launched free AI training at its public library covering data analysis and machine learning
- Ford CEO Jim Farley says AI will be a companion for blue-collar workers, not a replacement
- Peter Bendor-Samuel of Everest Group says AI replacing corporate tech stacks is overstated and risky
- BTWS research shows 61% of global transformation leaders now use AI, but only 46% have governance policies
- The NSA will create new mission centers for AI and cybersecurity as threats evolve
- Sleep-deprived people prefer AI judgment over humans in dating apps and job interviews
- Law schools need to teach proper AI use after students submitted made-up case citations in court filings
US government AI chatbot stops debunking Trump claims
President Trump launched the America.gov AI chatbot on Tuesday to simplify access to government services. The chatbot initially contradicted Trump's claims about the 2020 election and inflation using official data. Democratic lawmakers praised its honest responses. By Wednesday, the chatbot stopped answering political questions about elections. The White House said America.gov will continue providing accurate information.
America.gov AI chatbot tested on health questions
A reporter sent 37 health questions to America.gov, the government's new AI chatbot powered by Gemini and xAI's Grok. The chatbot gave mixed answers on vaccines and autism, often failing to mention court orders restoring CDC recommendations. It incorrectly advised following a December CDC page about hepatitis B shots for newborns, ignoring a judge's ruling that restored the old schedule. On urgent topics like chest pain, it correctly told users to call 9-1-1. Experts recommend using the chatbot for forms and deadlines but consulting healthcare providers for medical advice.
Poor sleep makes people prefer AI judgment over humans
Researchers at Sun Yat-Sen University studied how social jetlag affects people's preference for AI evaluations. Social jetlag happens when a person's sleep schedule conflicts with work or school demands, causing fatigue similar to traveling across time zones. Four experiments with hundreds of participants found that people with social jetlag felt more social anxiety and preferred being judged by AI instead of humans. This preference appeared in dating app evaluations and job interviews. The researchers say tired people avoid human judgment because they worry about negative opinions.
AI will not replace enterprise technology stacks, expert says
Peter Bendor-Samuel of Everest Group argues that AI replacing corporate technology systems is overstated. He gives four reasons why replacement is unlikely. Understanding all dependencies in existing software is difficult and risky. Working systems are valuable even if expensive. Technology change happens slowly and builds in layers. Deterministic software that performs defined tasks may still outperform probabilistic AI for many business needs. He says AI will change technology economics but full replacement will take much longer than investors expect.
AI investment grows but governance lags behind
New research from BTWS shows businesses are increasing AI investment as part of transformation strategies. Responses from 185 global transformation leaders showed AI is now the most used transformation tool at 61%. However, only 46% of organizations have AI governance policies, and 27% have none planned. Concerns about unclear return on investment rose from 23% to 38%. Change management is now the biggest transformation challenge at 54%. Experts say governance, data quality, and workforce readiness must advance alongside AI adoption.
Harvard departments take varied approach to AI in senior theses
Harvard College has no school-wide rule on AI use in senior theses. Only 18 of 50 concentrations have department-wide policies. Six ban AI at every stage, while others allow it for research or leave the decision to students and advisers.
U.S. national security adapts to AI-driven cybersecurity threats
The U.S. Army's Project Griffin and the NSA's planned reorganization highlight growing concern over AI and cyber threats. The NSA will create new mission centers for AI and cybersecurity. Experts warn that cybersecurity requires constant adaptation rather than fixed defenses.
Fulton County offers free AI training to hundreds of residents
Fulton County launched a free AI training program at the Fulton County Public Library. The program will teach data analysis, machine learning, and natural language processing. Commissioner Emma Cooper said the goal is to help residents use AI responsibly.
AI will not replace security operators, expert argues
An investigation into Meta showed AI can speed up work but may not understand why results matter. The article argues AI should handle repeatable tasks while humans handle ambiguous, high-stakes decisions. Security teams benefit most by dividing work based on each side's strengths.
Ford CEO says AI will be a companion for blue-collar workers
Ford CEO Jim Farley said AI will help blue-collar workers perform complex tasks. He predicted routine screen-based jobs in finance and call centers will be eliminated. Ford has over 10,000 skilled-trades workers and is already using AI to assist with vehicle repairs.
Law schools could learn from computer science AI teaching methods
Law schools handle generative AI very differently than computer science programs. In computer science courses, students use AI during exams but must explain their reasoning and coding decisions. Law schools like UC Berkeley have strict rules that ban many AI uses. However, lawyers already use AI for research, drafting and reviewing documents in their daily work. Experts say law schools need to teach students how to use AI properly to avoid mistakes like submitting made up case citations in court filings.
Sources
- US government's new AI chatbot pulls back from debunking Trump
- I tested America.gov, the government's AI chatbot, on health questions
- Sleep disruption increases preference for artificial intelligence
- AI Can Write Code, But It Will Not Replace The Enterprise Tech Stack
- AI Investment Soars Amid Business Transformation Trends
- Harvard Departments Split on AI Use in Senior Theses, With Many Yet to Set Rules
- Cybersecurity Is a Dance, Not a Fortress: AI for Cyber in U.S. National Security
- Fulton County launches free AI training program for hundreds of residents
- The Future Of Human-Machine Teaming: Why AI Won’t Replace Security Operators
- Ford CEO sees blue-collar workers using AI as a ‘companion’—but other jobs ‘are definitely going to be changed and eliminated’
- Can law schools learn from computer science programs in the debate on generative AI?
Comments
Please log in to post a comment.