By skyforbes 
I remember when GPT‑3 and 3.5 came out, I asked a lot of economics and political questions. They felt pretty economically liberal. For things like homelessness, housing shortages, and Milton’s quadrants, the answers leaned toward strong government intervention to fix societal problems, especially on the economic side.
Now, it gives more nuanced takes with multiple viewpoints and routes forward. I appreciate GPT‑5 for having more varied training‑data. The best part is rather than being decisive that there is only 'one correct answer', it'll explain why different people believe in different paths forward.
Although, it doesn’t feel like as big a jump from 4o as it was hyped, but the balance is better, so Sam Altman's goal of neutrality does seem to be getting closer..