Study: AI had limited effect in 2025 election but may impact future ones

Ar­ti­fi­cial in­tel­li­gence did not ap­pear to sig­nif­i­cant­ly in­flu­ence vot­er be­hav­iour or the out­come of Trinidad and To­ba­go’s 2025 gen­er­al elec­tion, but a re­searcher is warn­ing that with­out reg­u­la­tion, AI-gen­er­at­ed po­lit­i­cal con­tent could be­come a ma­jor threat by the next elec­tion cy­cle.

Uni­ver­si­ty of the West In­dies re­searcher Dr In­drani Bachan-Per­sad says while tra­di­tion­al fac­tors such as eco­nom­ic con­di­tions, gov­ern­ment per­for­mance and par­ty loy­al­ty re­mained the main dri­vers of vot­er de­ci­sions, AI is al­ready chang­ing po­lit­i­cal cam­paigns.

Her study, The Chal­lenge of Mis­in­for­ma­tion through AI-gen­er­at­ed Memes and Deep­fakes on So­cial Me­dia Plat­forms Dur­ing Elec­tions in Trinidad and To­ba­go, ex­am­ined AI-gen­er­at­ed con­tent used dur­ing the 2025 elec­tion, in­clud­ing memes, deep­fakes, syn­thet­ic au­dio and videos.

She found AI was used strate­gi­cal­ly in neg­a­tive cam­paign­ing to tar­get po­lit­i­cal lead­ers and con­tentious elec­tion is­sues. How­ev­er, she said the con­tent was large­ly dis­missed by vot­ers be­cause it was viewed through Trinidad and To­ba­go’s cul­ture of hu­mour, satire and pi­cong.

Dr Bachan-Per­sad warned this may change as the tech­nol­o­gy be­comes more ad­vanced.

She said by the next elec­tion, it could be­come much hard­er for cit­i­zens to dis­tin­guish be­tween re­al and fab­ri­cat­ed con­tent, cre­at­ing op­por­tu­ni­ties for mis­in­for­ma­tion and ma­nip­u­la­tion.

The re­search, pre­sent­ed dur­ing UWI’s In­ter­na­tion­al Me­dia and Com­mu­ni­ca­tion Con­fer­ence, traced the evo­lu­tion of dig­i­tal cam­paign­ing from 2010 to 2025, show­ing a shift from da­ta-dri­ven tar­get­ing to the wide­spread use of AI tools ca­pa­ble of pro­duc­ing con­tent quick­ly and di­rect­ing mes­sages to spe­cif­ic au­di­ences.

Dr Bachan-Per­sad said Trinidad and To­ba­go is par­tic­u­lar­ly vul­ner­a­ble be­cause of high in­ter­net pen­e­tra­tion and heavy use of plat­forms such as Face­book, What­sApp and Tik­Tok.

She is call­ing for stronger reg­u­la­tions around AI use in po­lit­i­cal cam­paigns, greater dig­i­tal lit­er­a­cy and more trans­paren­cy in po­lit­i­cal ad­ver­tis­ing and vot­er da­ta use.

She said AI will like­ly play a larg­er role in fu­ture elec­tions, as­sist­ing with re­search, con­tent cre­ation and da­ta analy­sis, but could al­so fu­el more so­phis­ti­cat­ed mis­in­for­ma­tion cam­paigns.

Dr Bachan-Per­sad said proac­tive mea­sures are need­ed now to pro­tect the in­tegri­ty of fu­ture elec­tions.