story

The global effort to curb alcohol harm will see a wave of change from the rollout of artificial intelligence (AI), but how big will it be? Will it be a ripple, crashing breaker, tidal surge or shattering tsunami? Alcohol Review’s annual conference in March will try to find out, while showcasing many other key developments in the field. To read on, please, register for the event and newsletter, or the free event preview and log in.

The global effort to curb alcohol harm will see a wave of change from the rollout of artificial intelligence (AI), but how big will it be? Will it be a ripple, crashing breaker, tidal surge or shattering tsunami? Alcohol Review’s annual conference in March will try to find out, while showcasing many other key developments in the field.

The economic energy has been given for massive change, with tech firms this year stumping up $1.5 trillion on an AI loss-leader, more than the yearly economic output of Indonesia. It seems inevitable that AI will spread further into alcohol research, advice, treatment, health promotion, alcohol marketing and law enforcement, among other areas.

More than few questions come to mind: Where is AI best suited to play a role and what are its limitations? What are the risks? How sustainable is it? AI is currently cheap, but might it become too expensive after the loss-making rollout splurge? Where is human expertise and insight needed? Will AI help to amplify or diminish human contributions? 

I will take my own trade as a starting point on some of this, with journalism requiring a rag-bag of different skills including: research, fact checking, storytelling and interviewing. We first see that AI is very good at scraping the internet for relevant research and writing readable, well-sourced summaries of its findings. And it does this extremely fast, without the typos or drafting errors typical of its human rivals. 

But AI falls short. It presents information which does not belong together, not recognising inconsistency. This is because it knows nothing about the real world its output seems to be about. AI is not sensitive to conflicting interests, misinformation, unintended ambiguities and downright lies. This means AI will always be a vector for unreliable alcohol information where safeguards are absent.

A lack of worldly understanding means AI does not and cannot lay out information as a story, which is an essential molecule of  human communication. Instead it writes impressively comprehensive lists. It does not how the information fits together, what is surprising, or what is contradictory, touching, funny or likely interesting to readers. AI, then, cannot lead us through the material with an engaging story. This severely limits its communication capabilities.

To tell a story one needs to develop an angle for the story which provides a hook to engage readers. It might be picking out a particular detail, presenting a new relevant piece of information, a new viewpoint or a human experience we can empathise with. This all requires some understanding of the real world and the people in it and what they care about. AI images also lack this crucial storytelling element.

AI, at least in its current form, does not add new information to what it serves up. It does not talk to experts or pull in new sources of information or opinion. It does not interrogate a database or get out knocking on metaphorical doors looking for quotes. And its summaries do not bring new images or metaphors to clarify a subject. Journalists add, but AI simply compiles what is on the web already. 

At some point AI might become more pro-active in their information searches. What it can already do is impressive. In an AI speech interview I found it accurately picked up what I was saying and formed questions from it. This is better than many human interviewers, who often do not prepare, listen or adapt their questions to the answers given. AI can definitely collect new information, but it seems to be more like a questionnaire than an interview.

Again AI is missing a key ingredient of what a journalistic enquiry does. Interviewees are chosen based on an understanding of the real world, including an understanding of their role, interests and expertise in a topic. They are persuaded to talk by instilling trust. And a line of enquiry should change as new facts are added, again requiring some knowledge of how the world works. Investigating alcohol harm is as challenging, if not more, than many other areas.

We should be open to the benefits AI may offer, but surely empathy and understanding will always be core to understanding human foibles?  AI seems set to fall short in crucial areas of journalism’s attempt to do this. And it seems likely these shortcomings will also hamper AI in other complex truth-seeking and communication tasks, like that needed to reduce alcohol harm. 

Note: For more on joining these online sessions on AI and the wide-ranging participant presentations, please visit the conference homepage and check out the new preview. ■

AR2026: Alcohol harm and artificial intelligence

+ all the latest research, advocacy and innovation.
Thursday, March 26th 2026, everywhere. Registrations open

Alcohol deaths in England were 32% above pre-pandemic levels in 2024, despite a 7% fall from the post-pandemic peak the year before, according to new new figures. A total of 38,399 people in England died of causes directly linked to alcohol drinking in the five years to 2024, over 9,304 more than if alcohol-specific deaths had remained at 2019 levels.■

UPDATED: The UK Chancellor Rachel Reeves this week gave respite from years of real terms alcohol tax cuts in her Budget statement, but made little headway on delivering on a manifesto promise to prioritise the prevention of health problems.

“I will support the great British pub through our new national licensing framework, encouraging councils to back our pubs and back late night venues with greater freedoms,” Reeves said in her yearly financial statement on Wednesday.

Reeves was referring to a much-criticised plan to change local alcohol licensing law to allow councils to put economic growth over health criteria. The government seems to be rewriting its proposals following an outcry from public health.

Councils are still being asked to “view licensing as part of their economic development or regeneration functions” , but other controversial aspects of the proposals have been “watered down or dropped”, posted James Nicholls, a critic of the proposals.

Reeves also said alcohol tax would remain the same in real terms next year. Health advocates recommend that the tax would need to rise in real terms to reduce record levels of harm.

But health advocates see this year’s tax freeze as relatively good news after years of real-terms tax cuts. “It is a clear signal that ministers aren’t bowing to a barrage of misinformation and aggressive lobbying,” said Katherine Severi of the Institute Of Alcohol Studies.

The Alcohol Health Alliance UK also welcomed the freeze, but it also repeated its call for a sustained increase of 2% above inflation and to develop an alcohol strategy.

It is a “welcome step forward”, said Alcohol Change UK, but added that it “expected bolder action” while also raising concerns about licensing reform, especially on alcohol delivery. ■

Less drinking among younger Australians has not outweighed harm within older cohorts, explains Professor Nicole Lee of the NGO Hello Sunday Morning and 360 Edge consutancy, who goes on to outline some of the applications and limits of artificial intelligence.

The Trump administration has taken a chainsaw to efforts to reduce alcohol harm. Here Mike Marshall, CEO of the US Alcohol Policy Alliance, helps make sense of the bewildering news flow.

“Strong alcohol policies are among the smartest investments you can make,” said WHO Regional Office for Europe and the WHO’s International Agency for Research on Cancer today at a two-day event launching two Handbooks on Alcohol Policy and Cancer Prevention.

“You can prevent more cancers with a taxation increase than you can with advice from doctors. If you want to prevent cancers you have to think about population policies in the long run,” said Jurgen Rehm, from Canada’s Centre for Addiction and Mental Health (pictured).

“The WHO European Region, and especially countries of the EU, are paying too high a price for alcohol in preventable cancers and broken families, as well as costing billions to taxpayers,” said Dr Gundo Weiler Director for Prevention and Health Promotion at WHO/Europe.

The pair recommend governments: increase taxes and minimum prices; raise minimum alcohol purchase age or drinking ages; reduce alcohol outlet density or opening hours; implement bans on alcohol marketing; create government monopolies to control alcohol sales.

The IARC Handbook Volume 20A evaluates the impact of reducing or quitting consumption on cancer risk while Volume 20B focuses on alcohol policies to reduce consumption. ■