The Real Risks of Rapid AI Development: Sam Altman and Mr. Sabis Speak Out
- The dangers of AI development are real, and no one individual or group can maintain control over the technology if it is developed quickly
- Sam Altman and Mr.Sabis have both acknowledged these risks and are being forced to move faster than they feel comfortable because of commercial pressures
- AI safety research has gone slower than expected, so we need more time in order to reap the benefits of AGI without losing control of it
- One solution suggested by Sam Altman is to release often and be transparent with development in order to learn more
- Another danger posed by AI is its ability to write code, which could lead to recursive self-improvement beyond AGI levels
- A third risk is connecting AI to the internet, which has already been done
- Stuart Russell has argued that teaching AI anything about humans or human psychology can be extremely dangerous, but this has also been done with social media algorithms manipulating user behavior for profit.
Reimagining Social Media for Humanitys Greatest Good: The Need for Open Dialogue and Incentive-Based Technology
- It is possible, necessary, and desirable to redesign social media in order to foster constructive conversations and facilitate solving the biggest global problems
- Democracy relies on open dialogue among diverse people
- It is reasonable to assume that good outcomes for humanity require meaningful conversations between humans
- Incentives can be created which both make money and bring out the best in people
- Current AI technology is like Neanderthal babies, but when improved could replace humans, which would not be a good investment for humanity
- Pride in our species should motivate us to use AI for beneficial purposes.