Trump vs Harris: Final AI Prediction
The final day is here. Within next 12 hours we will have a good idea on who the next US President will be (hopefully).
As mentioned earlier in my posts, we've built an AI model to predict the elections outcome to test how accurately we could predict the results from open source data. Our model’s greatest strength is also its greatest weakness, which is that it is entirely real-time and hence super sensitive to data inputs.
That said, in the last few days leading up to the elections, as hard as it is for us humans to navigate and trust the polarising information on who is winning and losing given the noise and propaganda around the elections, it is equally difficult for AI to make sense of the socio-political space and cancel noise.
Disclaimer: There is an important variable that I want to issue as a disclaimer when predicting election outcomes: the “invisible voter” that is undetected by polls. Most polls were wrong in 2016 about Trump because of this invisible majority that was voting Trump but was not comfortable to air their support publicly, and hence, went completely undetected by the traditional polling methods as Trump was a non-traditional candidate with no defined base. The tables in 2024 have now turned, with Trump's base very loud and quantifiable while there is a lot of 'unknowns' around Kamala. To check for confirmation bias in 2024 we must ask if Kamala could be a wild card with a defined Democrat base but also an additional ‘undefined’ and 'invisible' base that both traditional and non-traditional polling is unable to capture given the short interval between her candidacy announcement and elections? I have tried every effort over the months to identify, and capture this base for AI model but to what degree was it successful, I guess we will know in a few hours. If we go by data, momentum and public pulse, Trump is likely to win. If we factor in the 'invisible' element then anyone's guess is better than anyone's guess.
Here is our model’s final day results which are consistent with varying degree over the last 2 months:
National Level Prediction:
Donald Trump: 55%
Kamala Harris: 45%
Swing States Predictions:
Arizona: Trump: 52% | Harris: 48%
Georgia: Trump: 54% | Harris: 46%
Michigan: Trump: 50% | Harris: 50% (toss up)
Nevada: Trump: 51% | Harris: 49%
North Carolina: Trump: 56% | Harris: 44%
Pennsylvania: Trump: 53% | Harris: 47%
Wisconsin: Trump: 52% | Harris: 48%