Images of Donald Trump arrested by police officers on the streets of New York. A video that shows the apocalyptic future of the United States if Joe Biden wins a second presidential term. A statement by King Felipe VI apologizing to the Catalans for not having been neutral in the 2017 independence referendum. A dialogue between a Mexican candidate and Benjamin Franklin about the border situation between their countries.
All these contents have two things in common: they have been used officially by candidates or parties in campaigns or in political messages during this year and they were all made with artificial intelligence (AI). It is a small sample of the impact that this generative technology is beginning to have in the political conversation and that, surely, will increase in Latin America, as the presidential elections in Argentina and Ecuador approach, the second round in Guatemala, the regional elections in Colombia and the internal processes to elect presidential candidates in Mexico. All this will happen before the end of 2023.
Will these be the AI elections or will we still have to wait to see the potential scope of this technology? Will its impact continue to be limited to flashy videos that are broadcast on social networks officially or unofficially or will we see new proposals?
To answer these questions, it is necessary to evaluate the current extent of the use of AI in Latin America and it is interesting to review the data from a report recently published by HelloSafe. According to the study, the challenge of the pandemic led the private sector in the region to start betting heavily on AI from 2021. Half of Colombian companies already use these tools, as do 49% of Peruvian companies, 41% of Argentine and Brazilian companies, and 40% of Mexican companies. It is striking that 6 out of 10 companies that have incorporated this technology have done so for the areas of marketing and development of services such as contact center automation. That is, in functions that directly or indirectly require communication with the client.
Taking these data into consideration, and in regard to aspects that directly impact the electorate, the use of AI could also be expected to improve chats that allow party and candidate proposals to be known and that try to persuade them to support them. It is clear that the innovations will not be homogeneous throughout the region. Most of the advances are concentrated in countries such as Colombia, Brazil, Mexico, Argentina and Chile, where Internet penetration and the influence of digital communication is deeper than in other territories.
internal change
So far we have seen examples of elements that directly impact the voter, but AI also brings changes in the dynamics and working methods of campaigns that will have important consequences.
Next year we will surely see great advances in this same direction during the presidential elections of the United States. There, the Democratic Party and the Republican are already applying these tools to simplify arduous and complex data mining tasks. The goal is to better and faster identify donors and voters and discover patterns of behavior that can be exploited to drive donations and support. And just as Donald Trump’s campaign revolutionized the microtargeting with the questionable advice of Cambridge Analytica in 2016, very soon we will see how far the new tools can go.
But beyond data analytics, AI can also transform campaign teams by automating essential, time-consuming tasks like creating large volumes of social media posts or speech proposals. ChatGPT is capable of proposing a draft of a short speech in just seconds. Obviously, the results are not ideal. They are depersonalized, contain significant biases, and lack humanity.
The disproportionate use and lack of care and refinement of these tools can stifle the creativity and invention that characterizes human intelligence, but it cannot be denied that the agility they allow, if well calibrated, can help candidates with much smaller budgets and teams to more effectively face well-known and better financed figures in the future.
The ethical challenge
All these issues that we have mentioned carry a significant ethical challenge. How are you going to ensure that the deep fakes improved with AI do not feed hoaxes that serve to deceive the electorate? Will the new data mining practices represent a new challenge for the protection of personal information?
After the release of the Republican Party’s apocalyptic video of Biden’s second term, Democrats brought to Congress a bill to force the labeling of ads that are made with AI. That is a good first step, but surely there will be many more to take.
In communication -and especially in politics- the fundamental fight of our time is for our attention. It is the scarce good: a second, a minute, an hour, a day… cannot grow. Our attention span is extremely limited in the face of the enormous, overflowing and ultra-competitive supply of information that, directed by powerful audience fixation mechanisms thanks to sophisticated algorithms, competes for our time: the most precious and scarce asset.
In this hotly contested fight, AI has the ability to program itself to adapt with a perfect osmosis between our desires and behaviors and the information it can provide us. We went from asking questions to interacting with programs that reproduce sensations and artificial humanization environments. And offer us a comfortable uncritical and customized relationship capable of creating perfect bubbles.
In addition, AI has an enormous predictive capacity and can be used with solvency and depth to anticipate electoral behavior. This power can alter the distribution of resources and energies to concentrate all the strategic and persuasive efforts on those voters in dispute (who doubt whether or not to vote, or for whom); border (that can move between two nearby options); of transfer (that in electoral systems of second rounds they must vote for another option that was not their first choice).
Democracy and electoral processes in Latin America face extraordinary challenges generated by AI. As long as regulation does not adequately limit it, we are going to witness large and massive intervention experiments to condition or impose narratives and strategies. It is necessary to prevent AI from turning democracy into a laboratory for guinea pig citizens, where free will is replaced by the profound manipulation of wills and criteria.
