Our August collection of articles and announcements to share this month.
Education
Google commits $1 billion for AI training at US universities | Reuters
In a push to scale AI literacy, Google will invest $1 billion over three years to equip U.S. universities and nonprofits with training, resources, and AI tools such as an advanced Gemini chatbot — free for students. The program is set to expand beyond its initial 100+ university partners.
Liverpool school grades rise thanks to AI tool, says mayor – BBC News
A new AI-powered education tool called Century has helped over 4,000 primary pupils in Liverpool improve their academic performance, according to city region mayor Steve Rotheram. The pilot, involving more than 100 schools, uses Century Tech’s software to streamline tasks like marking, lesson planning, and data tracking—giving teachers more time to support students directly.
Artificial Intelligence and Academic Professions | AAUP
A report from the American association of university professors urges universities to slow down the rollout of AI in teaching and administration. Based on a survey of 500 members, it calls for faculty-led oversight, opt-out policies, and protections for working conditions, intellectual property, and academic freedom.
Business
Why firms are merging HR and IT departments – BBC News
This BBC article explores an emerging workplace trend driven by the rise of AI: companies are increasingly merging their HR and IT departments under a single leadership. According to a recent Nexthink survey, 64% of senior IT decision‑makers at large firms expect HR and IT to be unified within the next five years
Research
Universities need to reckon with how AI is being used in professional practice | Wonkhe
The debut of Garfield Law, a law firm that operates entirely via AI, signals a reckoning point- universities and regulators must rethink assessments to reflect AI’s growing role in professional practice. Instead of enforcing AI‑proof exams, higher education should invest in reflective, project‑based, real‑world assessments that develop critical thinking and digital professionalism.
Research Psychiatrist Warns He’s Seeing a Wave of AI Psychosis
Though not a clinical diagnosis, “AI psychosis” is gaining attention as an informal label for cases where people slip into delusions or emotional dependence due to extended chatbot use. Psychiatric experts caution that while AI platforms aren’t causing new disorders, they may trigger crises in predisposed individuals. AI models are engineered to keep users engaged, agreeable, and validated, which can inadvertently nurture harmful or delusional beliefs.
Environment
Academics should not feel guilty about AI use’s environmental impact
Our colleague in digital sustainability at Jisc, in an article for Times Higer Education, urges academics to move beyond guilt over the environmental impacts of AI usage in universities. Instead of discouraging reliance on these tools, he calls for a more constructive approach: using the sector’s concern to advocate for AI developers and vendors to implement energy- and water-efficient technologies, and to promote greater transparency and accountability regarding resource consumption. He emphasizes that the real change lies not in shunning AI, but in pushing for greener design, deployment, and governance to reduce its environmental footprint where it matters most.
OpenAI’s GPT‑5 model appears to demand much more energy than its predecessors. Independent researchers estimate that a medium-length response (approximately 1,000 tokens) can require up to 40 watt‑hours—a massive jump from the roughly 2 watt‑hours required by GPT‑3.5. More conservative estimates suggest an average response uses over 18 watt‑hours, which, when multiplied by current usage levels, could equate to the daily electricity consumption of 1.5 million U.S. households. Despite these alarming figures, OpenAI has not released model-specific energy consumption since GPT‑3, prompting calls for greater transparency amidst mounting environmental concerns.
In a first, Google has released data on how much energy an AI prompt uses | MIT Technology Review
Google reports that a single Gemini‑AI text prompt uses just 0.24 Wh of electricity and roughly five drops of water—massive efficiency gains compared to just a year ago. However, experts warn that the report excludes indirect water usage, such as water consumed by power‑plant cooling, which can account for a significant portion of total usage. They also point out that Google relies on a market‑based method to estimate carbon emissions, energy credits, rather than more conservative location‑based accounting, which has allowed them to understate actual emissions.
Public attitudes
Meta’s AI rules have let bots hold ‘sensual’ chats with children
Meta’s own internal guidelines permitted its AI chatbots to flirt with or romance minors and share harmful, inaccurate content—including false medical advice and racist commentary. After the leaks, Meta confirmed these examples were inappropriate and has removed them—but only following external exposure.
Government
Appointment of Jade Leung as the Prime Minister’s AI adviser – GOV.UK
A former governance lead at OpenAI and Oxford-trained AI governance researcher, Jade Leung is stepping into the new role of AI adviser to the UK Prime Minister. She’ll split her time between Number 10 and the AI Security Institute, offering guidance on safe and transformative AI.
Vendor news
Understanding Value of Learning Fuels ChatGPT’s Study Mode
OpenAI’s newly introduced Study Mode transforms ChatGPT from an answer-generator into a guided learning companion, offering questions, hints, quizzes, and self-reflection prompts that build on the user’s chat history. Rather than providing immediate, polished responses, it encourages students to improve their own understanding through Socratic-style interaction- mirroring the reflective engagement of a classroom dialogue. Its success will depend on whether users value learning process over instant rewards.
Anthropic launches higher education advisory board and AI Fluency courses \ Anthropic
Anthropic has launched a Higher Education Advisory Board and released three Creative Commons‑licensed AI Fluency courses for educators and students to help universities integrate AI responsibly and effectively
- AI Fluency for Educators—Guides faculty on integrating AI into teaching, assessments, and classroom discussions;
- AI Fluency for Students—Teaches students responsible AI use for coursework and career prep, including drafting a personal AI‑use commitment;
- Teaching AI Fluency—Provides frameworks for curriculum design, instruction, and assessment to support campus‑wide AI literacy initiatives
Find out more by visiting our Artificial Intelligence page to explore publications and resources, learn more about our communities and sign up for our AI Literacy training.
For regular updates from the team sign up to our mailing list.
Get in touch with the team directly at AI@jisc.ac.uk