Sam Altman believes that society is” close” to developing a superintelligence, an artificial intelligence with greater-than-human skills. He thinks that this will have a significant impact on the 2030s, including significantly lowering the cost of AI as computers take over the development of its system.
The CEO of OpenAI has outlined his plans for the remainder of this generation and the following in a blogging article posted on his website. While AI will result in job losses, Altman acknowledges that society will adapt and finally gain from novel scientific discoveries abroad.
In his final blog article, Altman outlines two important safety issues that need to be addressed: ensuring that superintelligence is widely available more than confined to a select few.
OpenAI, the father of ChatGPT, stated in January that developing superintelligence would be its main priority for the upcoming season. According to Altman,” We have just built networks that are better than persons in many ways.”
This implies that more people will be able to achieve things than they have ever before over the next ten and a half, and that possible only grows as AI expands past superintelligent levels. We are about to discover how much beyond human-level brains we may go, according to Altman.
AI may cost less and become even more affordable.
As the technology becomes more affordable, according to Altman, exposure to AI will increase significantly in the upcoming years. This value is attributable to AI’s ability to expand its own advancement by researching new processing substrates, more effective algorithms, and other facilities improvements. In the future, robots could be used to construct that equipment, lowering the cost of relying on people labor.
” If we need to create the first million human computers in the traditional manner, then they can run businesses, mine minerals, and do it all the time,” he said. The rate of progress will certainly be very different if we want to create more drones, which can also mean we can create more device processing services, data centres, etc., Altman said. The cost of intelligence may ultimately converge to nearly the cost of electricity as infrastructure production becomes automated.
The CEO added that the typical ChatGPT keyword uses about 0.34 watt-hours of electricity, which is almost the same as turning on a high-efficiency light for a few minutes and one-fifteenth of a teaspoonful of waters. However, according to a 2024 statement from the International Energy Agency, one ChatGPT fast used 10 terawatt-hours more energy annually than the annual total used for Google searches. Additionally, according to studies from The Washington Post, using GPT-4 to write a 100-word email requires about 519 milliliters of water.
There will be cultural advantages from AI, but some tasks will become redundant.
According to Altman, “whole groups of work” may dissipate as a result of improvements in AI in the 2030s. Numerous businesses, including Duolingo ( which has since clarified communications that AI will become replacing employees ), Salesforce, Shopify, and Klarna, have already taken advantage of the technology’s integration to reduce their staffing.
But, Altman claims that “we’ll be able to really think about innovative policy ideas we’ve never could have imagined because the world will be getting so many richer but immediately.” He envisions supporting universal basic income through his Orb program, which you electronically verify identities and stop scams. A social safety net like this might help to lessen the bad socioeconomic effects of job losses.
If history is any guide, he said,” We will figure out new things to do and new things to want, and assimilate new tools quickly ( job change after the Industrial Revolution is a good recent example ).”
At least some people may have brain-computer interface.
According to the OpenAI CEO,” real high-bandwidth brain-computer interface” could be one of the technological breakthroughs that superintelligence will allow by 2035. While “at least some people will probably determine to “plug in,” he wrote, “many people may not decide to have such an implant.”
This brings to mind Neuralink, the company that created the questionable head implant that connects the human mind directly to digital systems, as Elon Musk does. Although the goal is to make people with neurological disorders and disabilities live better, there are still many social and security issues related to this technology.
Space colonization is another Musk-technological development that Altman hopes to realize within the next ten years. In 2026 or 2027, SpaceX intends to begin sending people to Mars.
But foremost, we must address the health concerns.
As we advance and go beyond superintelligence, Altman recognizes the importance of addressing two key health issues. The primary concern is “alignment,” making sure AI systems consistently serve humanity’s long-term goals. He points to social media systems as an illustration of misaligned AI because they skilfully target short-term customer relationship while making people spend time in ways they may eventually dread.
The second problem involves ensuring that only a select few powerful people, businesses, or nations are able to control superintelligence. As well as tech giants like Google, Meta, Microsoft, and even OpenAI, the US and China are currently vying for AI dominance. Recently, the latter made the decision to stop operating entirely for profit, and it has since made repeated pledges to improve AI for all of humanity.
Altman doesn’t know whether or not these issues must be addressed before moving forward with superintelligence. He does acknowledge the difficulties by claiming that “given the economic effects,” it is crucial to widely distribute access to superintelligence, and that society should be able to form consensus norms and boundaries around its use.
Giving users a lot of freedom within the constraints of society, he wrote,” seems very important.” The better, the sooner the world can start to discuss these broad bounds and how we define collective alignment.
Read more about OpenAI’s potential partnership with Google and visit our sister website, eWeek, where we discuss Meta’s superintelligence lab.
