Artificial intelligence is coming for politicians
The use for artificial intelligence (AI) in the fabric of our economy and society is rapidly expanding. Because of this, it is worth considering the role this technology could play in the political sphere. With more than four billion people going to the polls this year, and following recent warnings from the World Economic Forum about the impact of AI-driven misinformation and disinformation on election outcomes, this feels all the more pertinent.
Eyes are already on AI to affect change in politics, but frequently in connection with better information feeds to enhance decision-making. There has not yet been sufficient focus on deploying AI where surely one of the quickest wins is waiting: calling out untruths, inconsistencies and evasive behaviour among senior politicians at the point of interview or when making public speeches.
The vision here is to nudge political operating standards towards those all professional service providers must contend with. Consider the likely reaction if hospital consultants, barristers or engineers were found to have deliberately lied or misrepresented facts, including risks, to patients or clients. At the very least, being struck off would be a very real prospect. By building upon certain existing business services with advances in AI, I believe politicians can be similarly held to account.
In the US, it is already possible to commission analysis to interpret the verbal and non-verbal behaviour of CEOs and other public figures when delivering earnings announcements and public statements. The service, valued by sophisticated institutional investors, tracks when communications are incomplete or lack transparency, along with other telling observations.
We can also draw inspiration from credit rating agencies, whose familiar designations of AAA (or equivalent), through to non-investment-grade territory like BB+ and below, determine the reception companies and countries will receive should they turn to the global bond markets for funding. Standard & Poor’s, one of the largest and most recognisable of the agencies, has set out a number of areas where it believes advances in AI can make a difference, from analysing large datasets to monitoring the digital footprints of rated entities, the latter in particular to divine early warning signals of credit deterioration.
Assessments on the nature of government and its structure are a key aspect of sovereign credit ratings. Here it is envisioned that it is possible to put in place a parallel/modified methodology to conduct analysis on key persons or parties in or contesting government. Although some fact checking already takes place in real time, with the main verification services it is only available on a lagged basis. But technology such as this could take the concept to the next level where the impact, as long as emanating from a credible source, would be acutely felt by those in public life.
Imagine a situation where, on broadcast media, AI is able to form a probability-based view on whether a politician, if not outright lying, is at least being overly evasive in their answers. It could count key omissions or attempts to side-step an awkward line of questioning. In response to questions beginning “how many”, which generally require a numerical answer, AI could detract marks from any politician replying by way of a counterattack. The same would apply to answers that contradict earlier statements and/or party positions, and most especially if the latter falls foul of economic orthodoxy. Most importantly, to enhance the value and impact, this process would play out in real time, although more detailed personalised reports showing trends over time could be made available.
It does not seem like a particularly high bar for sophisticated bespoke algorithms to collate and codify the results into some kind of aggregate score, including subcategories for honesty, consistency, even conviction. The results could then be communicated in a range of ways. On TV, we might imagine graphics displaying the outcome along the bottom of the screen. Radio hosts could verbally flag the highlights post interview. In the UK, an early triumph could come from Prime Minister’s Questions, simply by highlighting how many were satisfactorily answered while the PM is still on his feet in the house.
Almost certainly, the initial reaction from the worst offenders would be to swerve media programmes incorporating such tactics and without doubt the bots would be blamed for bias. The best counter is consistency of approach (industry-standard protocols applicable to all parties and individuals would need to emerge and be backed by relevant stakeholders) along with open architecture. Looking again to the rating agencies, they publish their methodology to fend off such claims, never more controversial than when they downgrade a sovereign.
A public fed up with being lied to, and on the back of the jarring experience of the pandemic and its handling, may well embrace tools to separate out the deceitful and devious from those in politics for the right reasons. Ambitious politicians might then choose to work only with the truth.
Ronan Donohue is founder and MD of Q4 Capital Advisors, and a former banker and capital markets consultant with a focus on FinTech and AI.
