CEO of OpenAI, who developed ``ChatGPT'' and ``DALL-E 2'', announces prospects for ``general-purpose artificial intelligence''

Conventional AI is a machine learning algorithm specialized to demonstrate performance in specific fields such as image generation, sentence generation, and speech analysis, but general-purpose artificial intelligence, so-called 'AGI' It is expected to be able to solve problems in various fields like humans by possessing various cognitive and information processing abilities. At the time of writing the article, AGI has not yet been realized, but Sam Altman, CEO of AI development organization OpenAI, announced the prospects for AGI development.

Planning for AGI and beyond

'Our mission is to ensure that AI, which is smarter than humans, benefits all of humanity,' said Altman, pointing out that it carries serious risks of misuse and social disruption. . Still, the benefits of AGI are so great that it cannot be stopped from developing, and society as a whole must figure out how to use AGI correctly.

And CEO Altman showed the following three as OpenAI's `` short-term prospects for AGI development ''.

First, as we approach the realization of AGI, AI developers should exercise caution in light of the society and circumstances in which AI is used. As AI advances into society, it is expected that it will face major problems such as how to combat prejudice and how to deal with unemployment. Altman says it is necessary to think.

Second, we should work on developing models that are more consistent and easier to control. Altman says that it is important for society to agree on how to use AI in a very wide range, and for individual users to have a lot of discretion within that range. For that reason, OpenaAI plans to limit the initial settings of the AI to be released considerably, so that users can easily adjust the operation according to their use. He also said that instead of suddenly releasing an AI that ``does everything well from the beginning,'' it is necessary to develop the technology little by little and continuously adapt to society.

Third, there should be a global dialogue to share the benefits, accessibility and governance of AGI broadly and fairly. At the time of writing the article, Altman said that OpenAI has the aspect of a non-profit organization, ``supporting other organizations to improve safety,'' and ``unlimited profits by including an upper limit on the return that shareholders can obtain. It is appealing that it aims to avoid the incentives to be obtained and the risk of deploying deadly dangerous technologies.

In addition, he proposed that it is important to have an independent audit before releasing a new system, and argued that it is necessary to set public standards for discontinuing development of AGI and the timing of model release. states that it is important for the world's major governments to have more than a certain level of insight into AI development.

In addition, as a ``long-term outlook,'' Altman said, ``The future of mankind should be decided by mankind. If the current speed of development is maintained, the world could become very different in the future, and the risks are immeasurable. Misadjusting AGI can cause serious harm to the world.'

Altman said, ``Moving to a world with super-intelligent AGI is probably the most important, hopeful, and terrifying project in human history. The downside is limitless, and hopefully it will unite all of humanity.We want to provide the world with an AGI that will bring prosperity to humanity in ways no one has yet imagined. I hope so,' he commented.

in Software, Posted by log1i_yk