The work of Costello et al. is a great example of how advances in artificial intelligence (AI) can be part of the solution to the problems and challenges that these same advances pose and magnify—in this case, the belief in misinformation and conspiracy theories.  

The experimental intervention by Costello et al. is really promising, especially considering its implementability and its short- and medium-term effect.  

However, it is important to emphasise the heterogeneity of the effect size of the intervention. Although the average effect is a 20% reduction in participants' expressed trust in conspiracy theories (as indicated in the press release), this effect depends on individual characteristics—such as the intensity of the person's belief in conspiracy theories, the importance of these beliefs for their worldview, and the trust the person places in AI and institutions. More specifically, among participants with higher belief in conspiracy theories, whose conspiracy beliefs are more important to their worldview, or with lower trust in AI and institutions, the effect size does not exceed 10%.  

In relative terms, a 10% reduction on a 0-100% confidence scale is not a dramatic change in some cases: for example, for a person who reduces his or her confidence from 95% to 85%, the intervention has not brought about a radical change in that person's conspiracy belief, as he or she still believes in the conspiracy theory. What the intervention has achieved (and deservedly so) is to cast some doubt on the original belief. This nuance has practical consequences when implementing this type of intervention in specific groups of the population, such as radicalised groups, where the expectation of effectiveness of the intervention should be lower and, depending on the goals and objectives of the intervention plan, insufficient in certain cases. 

In line with Costello et al.'s comments, I also believe that this research highlights the value of interventions based on the promotion of analytical, critical thinking and reflection to address beliefs in disinformation and conspiracy theories. However, I do not agree that this calls into question theoretical models that explain conspiracy beliefs as a result of the dissatisfaction of basic psychological needs (such as certainty, personal security, or the maintenance of a positive personal and social image). These models postulate that the satisfaction of these needs may help prevent or reduce the belief in conspiracy theories. Although there are some studies questioning this position, it is too early to rule out that interventions focused on meeting these psychological needs may contribute, along with other interventions such as the one proposed by Costello et al. to the complex and multifaceted solution of conspiracy theories.