In a new survey, 76% of scientists said that scaling large language models was "unlikely" or "very unlikely" to achieve AGI. When you purchase through links on our site, we may earn an affiliate ...