• Last time, we looked at AI and the labour market
• Automation displaces tasks, not whole jobs
• The graduate job market is rough right now
• AI raises worker productivity but may compress wages for some
• Today: a different angle. What does AI do to people, not jobs?
• Three themes: the attention economy, the environment, and mental health

Part 1: The attention economy

• “Your attention is the product”
• How recommendation algorithms work
• Filter bubbles: what the evidence says
• What companies are actually doing

Part 2: AI and the environment

• Energy and water costs
• Footprint in context
• AI as a climate tool

Part 3: AI and mental health

• The treatment gap
• Chatbots in therapy: what the evidence says
• Risks and hard limits
• Augmentation, not replacement

We will pause for discussion after each part

There are no right answers to the questions I put on screen. I’m curious what you think!

• Herbert Simon (1971): information-rich environments create attention-poor ones
• You have limited hours and limited focus. That’s the scarce resource
• The business model: capture attention, sell it to advertisers
• Google, Meta, TikTok. The service is free. You are the product
• Tristan Harris (ex-Google design ethicist): called it “the race to the bottom of the brainstem”
• The goal is not to inform you but to keep you on the screen
• Tim Wu’s The Attention Merchants (2016) traces this logic from 19th-century newspapers to social media

• Eli Pariser coined “filter bubble” in 2011
• The fear: algorithms seal you in a cocoon of confirming content
• Compelling idea. Two large experiments say: probably overstated
• Nyhan et al. (2023, Nature): reduced like-minded content on Facebook during the 2020 US election, polarisation barely changed
• Guess et al. (2023, Science): algorithmic vs. chronological feeds → little difference in political attitudes
• So filter bubbles may not drive polarisation the way we feared
• But that’s one outcome. Effects on anxiety, body image, attention spans, and sleep are separate questions, with worse answers

• In 2023, YouTube admitted that watch time alone is a poor proxy for satisfaction
• Started measuring “likes per impression” instead. Took over a decade
• The “outrage drives clicks” problem: content that makes you angry, anxious, or fearful gets shared more and watched longer
• The algorithm learns this. Surfaces more of it
• The Facebook Papers (2021, WSJ): internal research showed Instagram made body image worse for teenage girls
• Facebook had the data, but kept the feature because fixing it would have cost engagement
• The problem is structural: company revenue and user wellbeing point in opposite directions

• The sceptical read: most are opt-in, or reach a minority of users
• The core advertising model? Unchanged
• Also handy to point at when regulators come knocking
• US Surgeon General (2023): called for warning labels on social media for under-18s
• Amy Orben (Cambridge): transparency about why content is recommended changes behaviour more than usage caps
• Knowing you’re being nudged turns out to matter

• Training a large model costs serious energy. Companies rarely disclose the numbers
• Patterson et al. (2021): training GPT-3 produced ~502 tonnes of CO₂ (see table below)
• GPT-4? Nobody knows. Estimates range from 5x to 50x GPT-3
• But training is a one-off cost
• Running billions of queries (inference) is ongoing, and at scale it probably dwarfs training
• Microsoft’s carbon emissions rose 29% between 2020 and 2023, partly driven by AI
• Google’s 2024 Environmental Report: emissions up 48% since 2019, largely from data centres

• Energy gets the headlines. Water gets ignored
• Li et al. (2023), “Making AI Less Thirsty”: training GPT-3 used ~700,000 litres of freshwater for cooling
• A typical ChatGPT conversation: about 500ml, roughly a bottle of water
• Data centres get built where land and electricity are cheap, often in water-stressed regions: Phoenix, Las Vegas, northern Chile
• GPUs last 3-5 years. Manufacturing needs lithium, cobalt, and rare earth metals, each with its own environmental cost
• None of this shows up in the headline CO₂ figures

Current estimates (IEA, 2024):

• Right now, AI’s direct climate footprint is real but not huge compared to aviation or steel
• The worry is growth rate: AI electricity use grew ~60% year-on-year between 2022–2024
• Some projections: AI could use as much electricity as France by 2030
• The Jevons paradox: more efficient models → more usage → efficiency gains get cancelled out
• GPT-4 is far more efficient per query than GPT-3. But there are vastly more queries

The same technology that burns energy may also help cut emissions:

• GraphCast (DeepMind, 2023): 10-day weather forecasts, faster and more accurate than traditional models. Helps manage renewable grids
• Grid optimisation: AI predicts electricity demand, balances wind and solar in real time
• Wildfire prediction: better detection and spread modelling
• Data centre cooling: DeepMind’s RL system cut cooling energy by 40%
• AlphaFold: protein structure prediction, opening paths in climate biotech

Rolnick et al. (2022, ACM Computing Surveys) identified 70+ high-potential ML applications for climate change

• But: these are potential benefits, not guaranteed ones
• Many need serious compute to develop
• The teams building high-energy AI and the teams working on climate AI are mostly different people at different companies

• WHO estimate: 75% of people with mental health conditions in low- and middle-income countries receive no treatment at all
• Even in wealthy countries:
  • Average US wait for a first psychiatric appointment: 25 days nationally, 30+ weeks in rural areas
  • A therapy session costs $150–250 out of pocket
  • ~1 psychiatrist per 100,000 people in low-income countries
  • Stigma stops many from seeking help where care does exist
• This is the context for AI mental health tools
• If the alternative is literally nothing, the calculation looks different than if you’re imagining replacing well-funded human care

• Woebot (2017): first major CBT-based chatbot, 2M+ users. Follows clinical CBT protocols, not a free-form LLM
• Wysa: similar model, piloted by the UK NHS in several trusts
• Replika: not therapy, but an “AI companion.” People form real emotional attachments
  • In 2023, the company changed its relationship features without warning
  • Users reported grief and withdrawal symptoms
• Best recent evidence: Habicht et al. (2025, JMIR). Generative AI used alongside group therapy for real patients
• Improvements in clinical outcomes and patient engagement
• Key detail: AI was supplementing human therapists, not replacing them

• Opel and Breakspear (Science, 2026), two clinical neuroscientists, say AI “may reduce care inequities when deployed responsibly”
• That “may” and “responsibly” are doing a lot of work
• What the evidence contains:
  • Small RCTs: short-term symptom relief for mild-to-moderate anxiety and depression
  • Very few studies run past 8–12 weeks. Long-term effects? Unknown
  • High dropout: many people stop using these apps quickly
  • Publication bias: positive results get published; negative ones mostly don’t
• Real signals that AI can help some people with some conditions, short-term. But the evidence base is much thinner than the marketing suggests

These are documented, not hypothetical:

• Dependency. When Replika changed features in 2023, users reported grief comparable to losing a real relationship. What happens when an app you’re attached to gets discontinued?
• Data privacy. Therapy touches the most private parts of someone’s life. Chatbot data is rarely protected like clinical notes. Where does it go?
• Harmful responses. Moore et al. (2025): LLMs expressed stigmatising attitudes toward mental health conditions and gave harmful crisis responses
• No crisis escalation. A chatbot cannot call an ambulance, contact a GP, or do anything in the physical world
• Regulatory gap. Most mental health chatbots are classified as apps, not medical devices. No clinical regulation in the US, UK, or most of the EU. The EU AI Act (2024) classifies some AI health tools as high-risk, but most chatbots still slip through

But also consider:

• Therapy with a human isn’t fully private either: notes, supervision, insurance coding
• “Harmful responses from AI” vs. zero access to care for 75% of the world
• The strongest critique: AI tools may give governments and employers an excuse to avoid funding real care

LLMs cannot:

• Hold persistent memory between sessions. Every conversation starts fresh, the opposite of how therapy works
• Verify what you tell them. A therapist builds context over months
• Contact emergency services, a GP, or anyone in the physical world
• Provide legally enforceable confidentiality
• Diagnose or prescribe medication
• Read body language, tone, or facial expression

LLMs can:

• Be available at 3am
• Be patient. Not judge. Not get tired
• Deliver structured info on anxiety, depression, coping strategies
• Scale to millions at near-zero marginal cost

These are different capabilities, not better or worse. Whether what AI can do matches what people in distress actually need is the question worth asking.

The most credible use cases right now:

• First contact: AI handles initial triage, reduces stigma, connects people to professionals
• Between-session support: coping strategies and mood tracking between weekly appointments (the Habicht et al. model)
• Stepped care: AI for mild symptoms, human professionals for moderate-to-severe
• Admin burden: AI writes session notes, handles scheduling, finds local services. Frees therapist time for actual care
• OpenAI updated ChatGPT in 2025: added safety messaging, crisis hotline signposting, trauma-informed language. An acknowledgement that people already use the product this way
• Who decides what “responsible” looks like? Clinical psychologists train 7-10 years working with distressed people. App developers mostly don’t

Attention economy:

• Wu, T. (2016). The Attention Merchants. Knopf
• Nyhan, B. et al. (2023). Like-minded sources on Facebook are prevalent but not polarizing. Nature
• Guess, A. et al. (2023). How do social media feed algorithms affect attitudes and behavior in an election campaign? Science

Environment:

• Patterson, D. et al. (2021). Carbon Emissions and Large Neural Network Training. arXiv
• Li, P. et al. (2023). Making AI Less Thirsty. arXiv
• IEA (2024). Electricity 2024. International Energy Agency

Mental health:

• Habicht, J. et al. (2025). Closing the accessibility gap to mental health treatment with generative AI. JMIR
• Opel, N. & Breakspear, M. (2026). Harnessing AI for global mental health. Science
• Moore, D. et al. (2025). Stigmatising attitudes in LLMs. Crisis response study

General:

• Rolnick, D. et al. (2022). Tackling Climate Change with Machine Learning. ACM Computing Surveys
• 🎬 The Social Dilemma (Netflix, 2020)