The use of artificial intelligence (AI), fake profiles, and paid bloggers may have skewed the public’s perception of who is ahead in the April 28 General Election, but political analysts insist that while these tools may shape online chatter, they do not replace science.
They may not be willing to tell the public who is winning or what the ground is really saying just yet, but according to seasoned pollster and political scientist Derek Ramsamooj, AI does not trump data.
“In politics, the sophistication of political marketing via messaging would perhaps give an illusion of what the population is thinking. We have all seen where media can shape images by a headline, by pictures on the front page.”
“Likewise, now in the utilisation of AI, you can program the AI to say, ‘Use emotive language to show hurt in people or to show hope in people.’ But remember, that’s only in a social media space. The political momentum must be shown on the ground.”
Ramsamooj told Guardian Media yesterday that AI serves primarily as a tool for pushing political messages, not measuring public sentiment.
To truly understand voter inclinations, Ramsamooj stresses the importance of polling marginal constituencies. He points out that while the major parties’ strongholds may hold steady, the real deciders are the swing voters in key battlegrounds.
“So you look at the voter behaviour pattern in the percentage forms. So, if you won a constituency by eight per cent or ten per cent of the votes, then basically you’ve got to get polling divisions that are below that eight per cent or ten per cent and poll that particular electorate in that polling division.”
He said accurate polling must reflect a full cross-section of the population—factoring in ethnicity, age, gender, and socioeconomic status—if analysts hope to get a real read on the electorate.
Ramsamooj also notes a fundamental distinction between political polling and market research. He explains that voter decisions often remain fluid right up until election day, with choices shaped by campaign messaging and fast-moving developments.
Polls, he said, must not only reflect public opinion at a moment in time but also account for the ways that voter behaviour might shift as elections near.
However, the Council for Responsible Political Behaviour (CRPB), chaired by Dr Bishnu Ragoonath, is not primarily concerned with the use of AI or bots in political campaigns but the content and messaging conveyed through these tools.
The CRPB’s focus is on ensuring that political messaging adheres to ethical standards and does not promote divisiveness, hatred, or misinformation.
“The tool is being used, in most instances, in a negative way. Social media will publicise more of what we cannot verify in terms of whether it’s authentic or AI-generated. But the point about it is when they go on a platform and they talk, we could assess what is being said therein.”
Ragoonath acknowledged the difficulty in determining AI’s effect on voter sentiment, stating that it was challenging to assess whether AI-generated content swayed voters or not.
He also noted that traditional indicators, such as attendance at political meetings, similarly did not necessarily correlate with election outcomes.
Furthermore, the political scientist has observed that while core party supporters of the People’s National Movement (PNM) and United National Congress (UNC) were generally loyal, there were indications of dissatisfaction within these groups. This dissatisfaction, he said, could lead to reduced voter turnout or support for opposing parties, making it challenging to predict election results accurately.
There are 17 political parties contesting the general election in just over two weeks.
Since the date was announced by Prime Minister Stuart Young on March 18, both the PNM and UNC have accused the other of using AI to spread propaganda and misinformation.
