🔁Mario Nawfal:
🇺🇸 AI MODEL PASSES TURING TEST—FOOLS HUMANS 73% OF THE TIME
OpenAI’s GPT-4.5 just aced the imitation game.
In a new preprint study, researchers found that when given a “persona,” GPT-4.5 was mistaken for a real human 73% of the time—beating actual humans in some rounds.
That smashes the 50% benchmark of Alan Turing’s famed test for machine intelligence.
Without a persona? Just 36%. Meta’s LLaMa did similarly. OpenAI’s current GPT-4o? Only 21%.
The study suggests social engineering, job automation, and tech trust issues may soon become very real.
Source: Futurism
https://x.com/MarioNawfal/... https://x.com/MarioNawfal/...
🇺🇸 AI MODEL PASSES TURING TEST—FOOLS HUMANS 73% OF THE TIME
OpenAI’s GPT-4.5 just aced the imitation game.
In a new preprint study, researchers found that when given a “persona,” GPT-4.5 was mistaken for a real human 73% of the time—beating actual humans in some rounds.
That smashes the 50% benchmark of Alan Turing’s famed test for machine intelligence.
Without a persona? Just 36%. Meta’s LLaMa did similarly. OpenAI’s current GPT-4o? Only 21%.
The study suggests social engineering, job automation, and tech trust issues may soon become very real.
Source: Futurism
https://x.com/MarioNawfal/... https://x.com/MarioNawfal/...
3 months ago
