An artificial intelligence model built on historical voting patterns and current political data has produced a detailed hypothetical forecast for the 2028 U.S. presidential election, pitting Vice President JD Vance against California Governor Gavin Newsom in a general election contest. The projection, featured in a viral video from the Election Time YouTube channel, moves through state-by-state results and fills in an Electoral College map as polls close across the country. Before modeling the general election, the AI generated its own primary picks.
How the Simulation Was Built
The video’s creator asked ChatGPT to identify the likely Democratic and Republican nominees for 2028, then requested a detailed prediction of election night returns. On the Democratic side, the model chose Newsom, even though early polling shows him trailing former Vice President Kamala Harris by 15 percentage points. Prediction market Kalshi gives Newsom a 19 percent chance of winning the Democratic primary.
On the Republican side, ChatGPT selected Vance. He currently leads GOP primary polling and carries a 41 percent chance of securing the nomination, according to Kalshi odds.
Election Night, Hour by Hour
In the first wave of poll closings at 7 p.m. Eastern time, the model has Vance carrying Indiana and Kentucky, while Georgia tilts his way by 4 points. The AI describes this as a significant Republican improvement over 2024. After the early returns, Vance leads 44 electoral votes to 16.
By 8 p.m. Eastern, the forecast reaches the largest block of the night: 171 electoral votes across 16 states and Washington, D.C. In this window, Vance wins New Hampshire by 0.9 points, which would be the first Republican victory there since George W. Bush in 2000. Pennsylvania, North Carolina, and Florida also tilt Republican, pushing Vance’s total to 178 electoral votes.
The 9 p.m. hour brings two states the model calls historic flips. Minnesota leans Republican by 1.2 points, which would mark the first GOP win there since Richard Nixon in 1972. Wisconsin and Michigan both show narrow Republican leads.
Reaction and Regional Patterns
The video has spread widely on social media, renewing discussion about the 2028 race well before the contest itself. The final Electoral College tally of 326 votes for Vance and 212 for Newsom led observers to point out that Newsom’s projected showing would be the weakest Democratic performance since Michael Dukakis lost to George H.W. Bush in 1988.
The simulation’s regional results also depart sharply from recent history. The AI projects Vance winning Nevada and Arizona, along with several traditionally Democratic-leaning industrial states, all by narrow margins. Newsom holds the Pacific Coast and the Northeast, along with New York and Colorado.
What This Means
AI-generated election forecasts are speculative tools, and their output depends entirely on the assumptions and data behind them. ChatGPT’s prediction represents one possible outcome, shaped by how the model weighs historical voting data, demographic trends, and current political indicators. Real results hinge on turnout, candidate performance, unforeseen events, and shifts in voter behavior, none of which a simulation can predict with certainty.
The exercise shows both the capability and the limits of large language models in political analysis. The video offers a detailed look at how one AI system might model an election scenario, but actual 2028 outcomes will depend on economic conditions, campaign dynamics, and voter registration patterns that remain uncertain for years to come.
