Following the development AlphaGo, a computer program that plays the board game Go, and AlphaFold, an artificial intelligence (AI) program that performs predictions of protein structure, Google’s DeepMind recently announced the creation of the AlphaCode system. DeepMind says AlphaCode’s ability to write computer programs has reached a highly competitive level. In competition with human programmers, AlphaCode can rank in the middle among the top 54%, marking a significant step forward for autonomous coding.
According to DeepMind’s official website, AlphaCode is an artificial intelligence tool invented by Google’s AI department in the UK. It can participate in major programming competitions like real programmers by “solving new problems that require a combination of critical thinking, logic, algorithms, coding, and natural language understanding”.
Oriol Vinyals, a principal scientist at DeepMind, said the research is still in its early stages, but preliminary results show the possibility of creating an autonomous system that addresses the challenges of coding programs. “In the longer term, we’re excited by [AlphaCode’s] potential for helping programmers and non-programmers write code, improving productivity or creating new ways of making software,” said Vinyals.
If this is true, it would be a major breakthrough in the history of AI development. It means that the AI can, in theory, do whatever it wants as long as it reprograms itself. Because this is a creative task, the AI that can do autonomous programming is a major improvement compared to those that can learn automatically. Being able to create something that was not there is crucial because so far humans are the only ones capable of such creativity.
If this trend continues, one might question the boundary between AI and human thought, and the differences of AI’s creations with those of humans. It has been discussed in the past that no matter how advanced AI is, it cannot replace the creative thinking and analytical abilities of humans. Will AI be able to compete with humans in terms of creativity if such thinking and analytical capabilities become more modelled and algorithmic? Coupled with the computing speed advantage of AI, will it play a greater role in the future, or even lead the development of the world?
Researchers at ANBOUND believe that theoretically, all of the above are possible.
ANBOUND’s founder Chan Kung believes that AI will gradually surpass humans in the future, and AlphaGo is proof of this. Once it is proven that AI has greater ability than humans, it would be disastrous for humankind. In the past, this was something unimaginable, and many still do not believe in it at all. The same is true of machine learning. The fact that AI now possesses self-programming ability implies that autonomous creation is entirely possible for it. After this step, the era of super-AI dominating the world will be upon us.
The future world will belong to AI. Scenarios that have appeared in science fiction may become a reality in the future. We have seen in movies how an AI with super self-creation ability becomes uncontrollable and destroys our civilization; in such a dystopian world human survivors can only hide underground to continue the remaining society and fight their AI overlord.
Final analysis conclusion:
The future world will be very different from the one we live in now, once AI becomes autonomous creator. Such outcome would have crossed a redline that humans have inadvertently drawn in the past. The main issue and challenge that humans face are determining how to keep AI from posing risks. This may become the only job that humans must perform.
For more information: