Christian Szegedy asked on X whether people would use an AI they considered “dumber” than themselves.
Two replies complicated the premise. Lewis Tunstall said he was surprised that “Yes” was not closer to 95%, adding that people in the “X bubble” had probably used GPT-3.5. That was Tunstall’s speculation, not a measured claim about GPT-3.5 use or a captured poll result.
Bojan Tunguz drew a distinction between intelligence and expertise. He wrote that, until the summer, AI systems were “dumber” than he was while still having expertise far beyond his own.
The exchange does not establish a consensus or define how intelligence should be measured.
—
Not ranked yet
—
Not ranked yet
