GENEVA — Scientists from a consortium of European and North American research universities convened this week at the International Science Communication Forum to address what they described as a widening gap between rapid advances in laboratory research and public understanding of those discoveries, calling for urgent reforms in how findings are reported, funded, and taught in schools.
The forum, held over three days at a lakeside conference center outside Geneva, drew more than 400 researchers, science journalists, educators, and policy advisers from 34 countries. Delegates debated topics ranging from the ethics of preprint publishing to the surge in artificial intelligence tools being used to accelerate experimental design, with several sessions drawing standing-room audiences well into the evening.
At the center of discussion was a landmark survey of 12,000 adults across 18 nations, released Monday by the Global Science Literacy Initiative, which found that fewer than one in three respondents could correctly describe how a randomized controlled trial works, and nearly half could not distinguish between correlation and causation in a hypothetical example. Researchers said the results were largely consistent across income levels and education systems, pointing to structural rather than individual shortcomings.
“We are producing more peer-reviewed science than at any point in human history, yet the mechanisms by which that science reaches ordinary people have barely changed in a generation,” said Dr. Maren Lofthus, a science communication researcher at the University of Oslo who co-authored the survey. “That is not a public failure. It is a systemic failure.”
Delegates heard presentations on a range of proposed remedies. One widely discussed model, piloted in schools across Finland and South Korea, embeds scientific reasoning into humanities and social studies classes rather than isolating it within dedicated science periods. Early data from the pilot, covering 6,200 students over two academic years, showed a 22 percent improvement in participants’ ability to evaluate the credibility of a scientific claim compared with control groups.
Other sessions examined the role of social media platforms in accelerating the spread of contested or retracted studies. A team from the University of Toronto presented an analysis of more than 800,000 posts referencing scientific papers and found that articles later flagged for methodological errors received, on average, 3.4 times more shares in the first 48 hours after publication than those that survived rigorous post-publication peer review. The researchers attributed the disparity partly to the counterintuitive nature of well-designed studies, which often produce more qualified or less dramatic conclusions than poorly constructed ones.
“The incentive structure rewards novelty and surprise,” said Dr. Jonah Percival, a computational social scientist at Toronto who led the study. “A study showing that a drug modestly reduces a risk factor in a specific subpopulation does not travel well on the internet. A study claiming a revolutionary breakthrough does, even if that breakthrough later collapses.”
Funding bodies came in for particular scrutiny during an afternoon plenary on incentive structures in academic publishing. Several delegates argued that grant evaluation panels, by heavily weighting publication counts and citation metrics, inadvertently reward the behaviour that produces unreliable science. A working group formed during the forum will draft a set of proposed alternative metrics for consideration by national science funding councils in Europe and North America, with a report expected within six months.
A separate session drew particular attention to the growing number of predatory journals accepting papers with little or no peer review, in exchange for publication fees that can run as high as several thousand dollars per article. Delegates from lower-income countries noted that researchers at under-resourced institutions are disproportionately targeted by such journals and may lack the training or mentorship to identify them, effectively creating a two-tier system in which geography and institutional affiliation determine whether a scientist’s work is rigorously evaluated before reaching the public domain.
Forum organisers said they would publish a joint declaration in the coming weeks outlining recommended standards for science communication, including voluntary guidelines for researchers on how to present uncertainty and confidence intervals to non-specialist audiences. Whether governments or funding bodies will adopt the recommendations remains to be seen, but several national science academies attending the forum indicated they would consider incorporating them into existing grant evaluation criteria.
The next session of the forum is scheduled for the following year in Nairobi, where organisers plan to expand participation from African and Latin American institutions, regions they acknowledged were underrepresented at this year’s gathering despite producing a growing share of global research output.