TechCast is proud to present results from our study on Promises and Perils of AI. This study was fortunate to include a synthesis of the best judgment from our experts and responses from ChatGPT … that fabled merger of humans and machines. We gently suggest this study may be a model for future decisions made jointly by both AI and humans.
The results lead us to conclude that AI is a powerful tool, but the bold claims and deep fears over a purported “singularity” may be overblown. This study is limited, but we find no evidence to support the prospect for an AGI that is superior to humans, mass unemployment and existential threats commonly professed to be all but certain. As we will show, the data clearly suggest that AI will become far more powerful, yet subject to humanity’s unique powers of higher-consciousness. Our results forecast moderate to strong capabilities for controlling the dangers of AI, only a modest rise in unemployment, and beneficial social and human impacts.
These conclusions challenge the current fears over the AI Revolution, although we readily admit to the possibility of being proven wrong. Because this study synthesizes expert and AI intelligence, it represents a small breakthrough in collective intelligence by including the intelligence of AI. TechCast is confident these findings could help dispel the dangers of a theoretical singularity and to plan for controlling this newfound power responsibly.
A quick summary of our findings:
Almost all 28 respondents in this study see no purpose to a moratorium on AI research; A halt would impossible to enforce and it would encourage rogue operators. It’s obvious.
While strong regulation is essential, we foresee moderate, and possibly even strong, capability to control AI, rather than the existential threats often foretold. That’s good news.
One of of our most salient findings dispels prevailing threats of mass unemployment. We confidently forecast modest increases of roughly 10-12% globally. More good news. Yet there is always the possibility of unemployment approaching crisis levels of 20% or more in some nations and industries.
Half of our respondents think present AI systems are close to AGI even now, and that a full-blown AGI is likely to emerge about 2037. The other half thinks that AGI will arrive “much later or never.” We think this means that AI will become far more intelligent, including simulations of human values, goals, emotions, purpose, etc. But only life possess consciousness, so humans will always prevail. In that sense, TechCast suggests that a super intelligent AI that surpasses human agency is not possible.
Finally, the data suggest substantial gains in social stability and prosperity as well as comparable advances in human intelligence and creativity. Maybe we could lighten up on all the “Sturm und Drang.”
Because this study compared our results with those from ChatGPT, we also gained useful insight into the limits of AI. ChatGPT answered only 3 out of our 6 questions, and its answers tend to favor the system itself. As our previous newsletters suggest, the greatest danger lies in the spread of errors, misinformation, and untold forms of subjective bias.
We conclude AI is likely to become another tool in humankind’s advancing ability to manage knowledge with intelligence, although vastly more powerful and provocative. But the widespread fear over super intelligence, mass unemployment and grave threats prophesied by advocates remains only a theory. Absent some unknown breakthrough, TechCast forecasts good control over AI, huge gains in productivity, improvements in society and gains in human creativity.
Details of the study and the expert comments are extensive, and the full report can be found here on TechCast’s newsletter.
