In Future Progress in Artificial Intelligence: A Survey of Expert Opinion, Vincent C. Müller and Nick Bostrom show how high-level machine intelligence and superintelligent artificial intelligence (AI) will evolve over the next few decades with a specific analysis on the impending risks for humanity.
Müller and Bostrom show that experts believe that high-level machine intelligence will be around by 2075, with a fifty percent chance that it will be around 2040-2050. Furthermore, they show that around one-third of experts believe that the development and progression of AI will be bad for humanity. To come to their conclusions, they conducted a questionnaire and invited 550 experts to answer it. These experts came from a pool of participants from conferences, members of specific associations, and reputable authors. Since it is hard to ask straightforward questions surrounding this topic, they had to come up with comprehensive definitions and simple questions to gather the best data. The questions included topics such as super intelligence, high-level machine intelligence, and personal questions. At the end of the article, the authors included the raw data and specific graphs that elucidated the results. Moreover, the authors did not come to strong conclusions at the end of the paper. Alternatively, they said, “we leave it to the reader to draw their own detailed conclusions from our results, perhaps after investigating the raw data. Let us stress, however, that the aim was to ‘gauge the perception’, not to get well-founded predictions.” Overall, the results from the questionnaire show how superintelligence is coming and that it could be bad for humanity. Because of this, the authors recommend that society better investigates the future and implications of superintelligence on humans.
The results from this study were interesting and reliable. The inclusion of limitations and how they selected the expert participants for the questionnaire contributed to the validity of the results. Furthermore, the presentation of the raw data was easy to understand and find patterns within, so I appreciated that. The fact that the authors let the readers come to their own conclusions was interesting. This proves that there are no definite answers surrounding the future of AI and superintelligence, but that it is still important to analyze the current trends and implications of such a phenomenon. Overall, I found this article reliable and would be interested to see this exact questionnaire distributed within 10 year intervals to see any similarities and/or differences in results.
I didn’t read this, but I thought that your analysis was really intriguing. You brought up a great point about how the reasoning behind the lack of definitive answers is because we have no idea what AI is going to look like and how it is going to affect society in the future. This is something we have touched on a lot in this class over the semester, and it does make me nervous to see just what technology and AI have in store for us in our lifetimes. When reading your analysis, I kept thinking about an article I had read earlier this semester, generally asking that even though we can, should we? As it pertains to technology and the advancement of AI. This is a very controversial topic, in my opinion, and I still don’t know where I stand on it necessarily. It is a little frightening to think that even experts don’t know what is in store for us in the technological realm.