Nick Bostrom Explains How The Advent of Human-Level Artificial Intelligence Might Not Last Very Long
Machine learning is currently advancing at a rapid rate. Recently Nick Bostrom looked at some current capabilities, and considered what longer term prospects of artificial intelligence mean for human society. The transition to the machine intelligence era is likely to have profound consequences for the planet.
At the recent Unexpected Sources of Inspiration (USI) conference, Nick Bostrom made a rather simple observation that no event, even of the order of the Second World War, has been able to hinder the linear demographic growth of the last two centuries (see the whole video below). From this point of view, it can be seen that there have been only two fundamental ethnological events in the history of humanity: the advent of agriculture and the industrial revolution.
Bostrom has headed up the Future of Humanity Institute, a research group of mathematicians, philosophers and scientists at Oxford University tasked with investigating the big picture for the human condition and its future. He has been referred to as one of the most important thinkers of our age.
His book Superintelligence: Paths, Dangers, Strategies
advances the ominous idea that “the first superintelligent machine is the last invention that man need ever make.”
Related articles
According to Bostrom, the agricultural revolution, marked the beginning of human activity in settling and domestication of the environment of Mesopotamia. The end of nomadism is at the origin of phenomenal changes such as the advent of private property, the emergence of social strata, the creation of state political structures, wars and a true market economy.The industrial revolution, a "gigantic historical anomaly" according to Bostrom, represents a paradigm shift in our relationship to wealth: while the economy could not previously cover the needs generated by population growth, the growth of wealth could now exceed that of just population groth for the first time.
The conviction of Bostrom in the light of past history is that artificial intelligence could be "the third fundamental event for Humanity" with an elevation of our species to a level of intelligence unimaginable to us today.
What is changing today is that the machine now seems to be exercising skills that have hitherto been judged to be only in the capability of humans: recognizing a movement, reading and analyzing a text, and creating links between different situations. AI Systems are now able to learn by themselves through experience or to create a piece of music, certainly without true sense as a whole, but coherent and harmonious to listen.
Bostrom reminds us, "We are now at a stage where technologies have already proved that they are efficient and useful, which encourages investment in this field."
Big challenges in terms of understanding, learning and reasoning for machines make the horizon still uncertain. According to a recent survey of AI experts, there is a 50% probability that superintelligence will emerge in 50 years and 100% in 100 years.
But for Bostrom, these stakes are very short term and finally close enough to the social choices discussed elsewhere. These are, he says, other issues with broader and more complex implications that should be our focus.
AlphaGo's victory in a broader perspective
 |
| Lee Sedol sits defeated by DeepMind's AlphaGo. Does a similar fate await our species? |
After the first match, Sedol commented, "I was very suprised because I didn't think that I would lose."
After the second match, he said, "I am quite speechless. I am in shock. I can admit that the third game is not going to be easy."
Conceding the challenge after the third match, Sedol admited, "I felt kind of powerless."
If and when human-level general AI is developed, the outcomes could be anecdotal to what Lee Sedol experienced in the narrow niche of an Asian board game. From October 2015, Sedol was extremely confident, and judging from the state of AlphaGo's development, the human would have probably bettered the AI. But just months later, AlphaGo's performance was superhuman."What this suggests is that the answer to this question, 'what happens if we achieve human-level AI is, I think, superintelligence," states Bostrom. "and possibly this happens shortly after we have human-level AI."
What are the possible scenarios with the emergence of super intelligence? What are the possible dangers and how to control a potentially hostile and more powerful force than yourself? What global governance of artificial intelligence should be introduced in this regard to prevent any misuse of technology?
Bostrom uses in this respect the myth of the King Midas who, having claimed from Dionysus the power to change what he touched in gold, comes to regret his wish so ardent when he embraces his daughter and transforms her into a statue of gold.
Because to want, by all means, to reach the ultimate goal of super-intelligence, we could very well see our creation turn against us.
Comments
Post a Comment
Thank you for your comment!