Science under suspicion: how real is the trust crisis?

It has become a buzzword among sci­entif­ic organ­isa­tions and polit­ic­al lead­ers: trust in sci­ence is said to be under­mined and endangered by grow­ing con­spir­acy the­or­ies and irra­tion­al­ity among the pop­u­la­tion, fuelled by the explo­sion of digit­al tech­no­logy. Is this fear jus­ti­fied? To sep­ar­ate fact from fic­tion, we inter­viewed Hugo Mer­ci­er, research dir­ect­or at the CNRS, part of the Evol­u­tion and Social Cog­ni­tion team at the Jean Nicod Insti­tute, and Ben Seyd, a polit­ic­al sci­ent­ist at the Uni­ver­sity of Kent.

It is not neces­sar­ily new – rather the cur­rent vis­ib­il­ity has been giv­en to it, as well as its integ­ra­tion into sci­ence policy, is. Con­cern gained momentum dur­ing the Cov­id-19 pan­dem­ic, against a back­drop of anti-vac­cin­a­tion and anti-lock­down move­ments. “This has res­ul­ted in a shift in aca­dem­ic focus: we moved from ‘know­ledge defi­cit’ mod­els – where the obstacle to the influ­ence of sci­ence is seen as rest­ing on insuf­fi­cient under­stand­ing among the pop­u­la­tion – to ‘trust defi­cit’ mod­els – where the obstacle is seen as the pub­lic’s neg­at­ive assess­ments of sci­ence and sci­ent­ists,” explains Ben Seyd.

Quant­it­at­ive data refutes the idea of a wide­spread trust crisis¹, 80% of French people trust sci­ence, mak­ing it one of the most trus­ted ‘organ­isa­tions’, just behind artis­ans and ahead of hos­pit­als, the army and the police.

A large-scale study pub­lished in Nature in 2025², cov­er­ing 68 coun­tries, also con­cludes that there is a high level of trust in sci­ence glob­ally: on aver­age, 3.62 on a scale of 1 to 5. The highest levels are observed in Egypt and India (over 4.26) and the low­est in Kaza­kh­stan and Albania (under 3.13). Accord­ing to this study, the level of trust in France is slightly below aver­age (3.43), which is not sur­pris­ing: in inter­na­tion­al com­par­is­ons, France gen­er­ally ranks at the lower end of the scale among highly edu­cated coun­tries in terms of inter­per­son­al trust or trust in institutions.

Not sig­ni­fic­antly. In France, the fig­ures sug­gest a cer­tain sta­bil­ity over time. A series of con­sist­ent data from 2001 to 2020³ estim­ated that 84% of French people said they had con­fid­ence in sci­ence in 2020, com­pared with 87–89% in pre­vi­ous waves. Data from the Sci­ences Po/OpinionWay baro­met­er cited above, which cov­ers the more recent peri­od, indic­ates a con­fid­ence rate of 78% in 2020 and 80% in 2025.

Else­where in the world, there is also a cer­tain con­sist­ency when lon­git­ud­in­al data is avail­able ⁴. “The United States has not seen a sig­ni­fic­ant decline in trust since the 1970s. In Bri­tain, it has even increased since 1997. There has been a dip since the Cov­id-19 pan­dem­ic, espe­cially in the United States, but it also affects oth­er groups: the mil­it­ary, reli­gious lead­ers, and so on,” explains Ben Seyd.

Over­all, yes. “When trust in sci­ence and sci­ent­ists is meas­ured togeth­er, the over­all dis­tri­bu­tions are sim­il­ar,” says Ben Seyd. How­ever, this wide­spread sup­port hides some nuances: a study pub­lished in 2024 in PLOS One⁵ showed, for example, that the level of trust var­ied greatly depend­ing on the dis­cip­line. It is sig­ni­fic­antly high­er for research­ers in the phys­ic­al and nat­ur­al sci­ences than for polit­ic­al sci­ent­ists, eco­nom­ists and sociologists.

Gen­er­ally speak­ing, no. How­ever, adher­ence to cer­tain spe­cif­ic the­or­ies, such as the causes of cli­mate change or the bene­fits of vac­cin­a­tion, can be cor­rel­ated with par­tis­an lean­ings. But a few coun­tries are excep­tions to this gen­er­al rule, with sig­ni­fic­ant polar­isa­tion of trust in sci­ence in gen­er­al accord­ing to polit­ic­al lean­ings. This is par­tic­u­larly true in the United States, where con­ser­vat­ives have had sig­ni­fic­antly less trust in sci­ence than lib­er­als since the 2000s⁶.

Accord­ing to Hugo Mer­ci­er, “it’s the num­ber of years of sci­ence edu­ca­tion. The more you’ve been exposed to sci­ence, the more you trust its res­ults⁷”.

The anti-vac­cin­a­tion move­ment has received a lot of media atten­tion, but “the num­ber of people who are strongly opposed to all vac­cines is actu­ally quite small: typ­ic­ally 2 to 5% of the pop­u­la­tion, depend­ing on the coun­try⁸. On the oth­er hand, a much lar­ger pro­por­tion of the pop­u­la­tion is some­what fickle when it comes to vac­cines. The papil­lo­mavir­us vac­cine, for example, is strug­gling to win people over in France,” explains Hugo Mercier.

Accord­ing to an IPSOS/EDF sur­vey con­duc­ted in 2025⁹, 33% of French people do not believe that the cli­mate is chan­ging (9%) or refute its human ori­gin (24%). These fig­ures are slightly below the glob­al aver­age: 37% of scep­tics, includ­ing 10% who deny cli­mate change and 27% who deny its human ori­gin. Accord­ing to the same sur­vey, “cli­mate scep­ti­cism is no longer on the rise” and “scep­tics are not bene­fit­ing from a cli­mate of opin­ion that is less favour­able to the envir­on­ment,” even in the United States.

There is no robust data on the evol­u­tion of con­spir­acy the­or­ising in France. How­ever, in 2022, a team of Amer­ic­an and Brit­ish research­ers ana­lysed¹⁰ the evol­u­tion of the pro­por­tion of Amer­ic­ans who believe in 46 con­spir­acy the­or­ies, with some obser­va­tions cov­er­ing nearly half a cen­tury. The research­ers repor­ted find­ing no evid­ence of a sig­ni­fic­ant increase in con­spir­acy the­or­ies over the peri­ods stud­ied, although the pop­u­la­tion believes oth­er­wise (73% of Amer­ic­ans think con­spir­acy the­or­ies are cur­rently “out of con­trol,” and 59% believe people are more likely to believe in con­spir­acy the­or­ies than they were 25 years ago).

Most of the time, no. Hugo Mer­ci­er and his col­leagues sought to under­stand the true nature of mis­trust¹¹ expressed by the 5 to 10% of people who say they do not trust sci­ence. They sur­veyed 792 Amer­ic­an par­ti­cipants about their trust in sci­ence in gen­er­al, their adher­ence to the sci­entif­ic con­sensus on basic know­ledge (the length of the Earth’s rota­tion around the Sun, the rel­at­ive size of the atom com­pared to the elec­tron, etc.) and their degree of belief in 10 con­spir­acy the­or­ies (there has been extra­ter­restri­al con­tact and it has been hid­den from the pub­lic, the Earth is flat, etc.).

Res­ult: adher­ence to basic sci­entif­ic know­ledge in the over­all sample is extremely high (over 95%), but it is also high among those who say they do not trust sci­ence or who fully sub­scribe to a con­spir­acy the­ory (over 87% in both cases). “Even flat-earthers agree with everything… except that the Earth is round,” com­ments Hugo Mercier.

This can some­times be explained by an under­es­tim­a­tion of the con­sensus. A study pub­lished in Nature in 2024¹² showed, for example, that emphas­ising the con­sensus on the human ori­gin of cli­mate change had a sig­ni­fic­ant, albeit slight, effect on accept­ance of this fact.

But for Hugo Mer­ci­er, when we reject a con­sensus, it is in most cases because it dis­turbs us. “The case of cli­mate change is typ­ic­al: accept­ing the sci­entif­ic con­sensus on its human ori­gin jeop­ard­ises cer­tain eco­nom­ic interests. What’s more, part of the pop­u­la­tion does not want to change its green­house gas-emit­ting beha­viour and, con­sist­ently, shows a cer­tain mis­trust of the exist­ence of glob­al warm­ing. But we can­not say that these people are irra­tion­al or will sys­tem­at­ic­ally adhere to oth­er untruths.”

Yes. The exper­i­ence of the Cov­id-19 vac­cin­a­tion com­mu­nic­a­tion cam­paigns is also enlight­en­ing. “Before the first wave of vac­cin­a­tions, nearly 80% of French people said they did not want to be vac­cin­ated. Shortly after the start of the vac­cin­a­tion cam­paign and the accom­pa­ny­ing com­mu­nic­a­tion, 80% wanted to be vac­cin­ated,” explains Hugo Mer­ci­er. The same res­ult was seen for HPV vac­cin­a­tion. “When aware­ness cam­paigns are car­ried out in schools, vac­cin­a­tion rates increase.”

Yes. “Trust in sci­ence can be partly explained by the ‘ration­al impres­sion’ mod­el: we trust sci­ent­ists more when we are impressed by the res­ults they achieve.” To reach this con­clu­sion, Hugo Mer­ci­er and his col­leagues presen­ted 696 Brit­ish par­ti­cipants with two texts describ­ing the pro­fes­sions of archae­olo­gist and ento­mo­lo­gist¹³. The first was peppered with impress­ive inform­a­tion such as “archae­olo­gists can now determ­ine the age at which a per­son who died tens of thou­sands of years ago stopped drink­ing their mother­’s milk, based on the com­pos­i­tion of their teeth.” The second was writ­ten in a neut­ral style. Expos­ure to the ‘impress­ive’ present­a­tions led par­ti­cipants to judge the sci­ent­ists involved as more com­pet­ent and their dis­cip­line more trust­worthy, des­pite almost imme­di­ately for­get­ting the spe­cif­ic con­tent of the texts.

“It’s dif­fi­cult to say at this stage,” says Hugo Mer­ci­er. How­ever, it could have unex­pec­ted poten­tial. An art­icle pub­lished in Nature in 2025¹⁴ shows that a dis­cus­sion with an LLM can increase vot­ing inten­tions by 10 points out of 100, a bet­ter res­ult than that obtained through video adverts. “It’s great­er effect­ive­ness is due to its abil­ity to finely tail­or argu­ments to indi­vidu­als’ ques­tions, some­thing that a simple mes­sage without dis­cus­sion can­not do.”

Interview by Anne Orliac