Can AI improve mental health support—or deepen existing inequalities?
- info4696689
- Jul 14
- 1 min read
I’ve just read Unequal Benefits, Unequal Harms by David Gilbert (link below), a thought-provoking discussion paper exploring the rise of AI-powered mental health chatbots and self-guided care.
The paper argues that AI has enormous potential to expand access to mental health support, particularly for people facing long waiting lists, affordability barriers, stigma, or social isolation. But it also raises critical questions about who benefits most—and who may be at greater risk of harm.
Some key themes that stood out:
🔹 Access vs. EquityAI tools could help bridge gaps in care, but there's growing concern that underserved groups may increasingly be left with digital support while others access human care.
🔹 Companionship vs. DependencyAs AI becomes more personalised and relational, there is a risk that users develop unhealthy reliance on systems designed to be available, validating, and engaging 24/7.
🔹 Personalisation vs. PrivacyThe promise of tailored support depends on access to deeply personal data, raising important questions around consent, governance, and commercial use.
🔹 Efficiency vs. Human ConnectionAI may help manage growing demand for mental health services, but can it replicate—or should it replace—the empathy, challenge, nuance, and relationship offered by human practitioners?
🔹 Innovation vs. BiasWithout careful design and oversight, AI risks reinforcing existing social inequalities and biases, particularly for marginalised communities.
The paper doesn't argue for or against AI in mental health. Instead, it asks a more important question:
"How do we ensure AI strengthens human care rather than becoming a substitute for it?"
As AI rapidly becomes part of health and wellbeing ecosystems, this feels like a conversation we need to have now—not after the technology is fully embedded.
Comments