All research

Misinformation, conspiracy beliefs, and AI

Who believes what, why — and what artificial intelligence changes.

Keep scrolling

This strand treats misinformation as a system rather than a personal failing: it asks who is susceptible and why, what the biography of a conspiracy believer looks like, and what happens when the machinery producing false claims stops being human.

Conspiracy belief has a biography

Following the same people across three decades shows that loneliness trajectories from adolescence onward predict conspiracist worldviews in midlife. This reframes conspiracy belief as something with developmental roots rather than a position solely adopted in adulthood, and points prevention toward adolescent social connection rather than adult fact-checking alone. Related work finds that perceiving your own group as a minority — even when it is factually the majority — is itself linked to conspiracy thinking.

Susceptibility is specific, not general

A programme of work identifies distinct phenotypes of susceptibility to true and false information, showing that being fooled by falsehood and rejecting truth are separable failures with different cognitive and personality profiles. Unusually, this extends to pharmacology: studies link both sertraline treatment and tobacco smoking to measurable shifts in susceptibility, treating credulity as a cognitive state with biological correlates rather than a fixed trait.

Using AI to study belief, and studying what AI does

Machine-learning analysis of online behaviour has been used to identify the psychological factors associated with conspiracy beliefs at a scale interviews cannot reach. The other direction matters more: work in Science warns that malicious AI swarms — coordinated, human-mimicking agents — represent a qualitatively new threat to democratic debate, because they manufacture the appearance of consensus rather than merely spreading claims.

What actually works

A scoping review of psychological interventions against misinformation on social media assesses what the evidence supports, and a meta-analysis establishes how conspiracy beliefs translated into real behaviour during COVID-19 — reduced social distancing, measured over time rather than at a single moment. The honest summary is that effects are modest but real, which is worth defending against both hype and dismissal.

  • 16Papers in this strand
  • 13Journals & publishers
  • 2020–Active since

Papers in this strand

Also relevant to

All publications

Related press coverage

Press Coverage

FAQ

Frequently asked questions

The forgotten side of cultural contact — including the majority's.

Next strand

Next strand

Acculturation and cultural change

Read more →