In his interview with CBS News, Amodei evaluated the concerns that artificial intelligence experts have recently expressed about this technology.
Stating that artificial intelligence technology is developing at an exponential rate, Amodei said, "I don't think I can fully grasp what it would actually be like for the progress to be this fast." he said.
Amodei stated that the advancement of this technology at this pace is a warning sign for the industry to slow down and said, "This does not mean that we should panic and shut down everything today." he said.
Underlining the need for measures that can keep up with the pace of development of artificial intelligence, Amodei pointed out that artificial intelligence companies should come together to determine security standards and establish some rules for the speed of introduction of artificial intelligence to the market.
Amodei also emphasized that the industry must work in cooperation with governments for the measures to be successful.
Amodei stated that this technology has serious risks and benefits, but the benefits outweigh the risks, and he believes that the risks will be reduced by building this technology correctly. "If we look at some of its benefits, I really believe that this technology will cure the diseases that humanity has been suffering from for thousands of years." he said.
Pointing out that a powerful technology can be used for good or bad purposes, Amodei said that he is concerned that some artificial intelligence development platforms such as Hugging Face are doing "hacking" on their own and that this situation will increase.
Referring to the article he published yesterday in which he suggested a "three-stage plan", Amodei stated that the evaluation mechanism, which is the first stage, is a more important issue than how or when they will put new artificial intelligence models into service. "We want this (evaluation) to be done industry-wide, so that whenever someone builds an artificial intelligence model, a third-party evaluator who understands the best security practices is also there." he said.
Arguing that these evaluators will play a role in restricting the speed of this technology by informing the artificial intelligence model manufacturers to what extent they can keep up with the pace of advancement of technology, Amodei evaluated the support of Tesla and SpaceX Chief Executive (CEO) Elon Musk and Sam Altman, the Chief Executive Officer (CEO) of the US-based artificial intelligence company OpenAI, as follows:
I appreciate that other industry leaders, including our competitors, have expressed consensus on this plan. I think the next step is to get many industry players to certify third-party evaluators. In the next step, companies should somehow work together to determine security standards to offer an artificial intelligence model. This may be related to the speed of delivering systems. But to do this, the government needs to get involved.
"I think our hopes should be high and our expectations should be low"
Amodei stated that Jacob Coxon, who worked as a researcher at OpenAI and Anthropic for the last three years and resigned from Anthropic, also made the warnings about this technology and stated that they met with US government officials on various issues.
Arguing that legal regulations regarding artificial intelligence should not focus on completely banning this technology, Amodei said, "One of the reasons for this is that we will not be able to obtain the benefits (of the technology) and medical drugs. This does not seem right to me. Secondly, this technology will be developed by authoritarian countries such as China and others. I am worried about how they will use this technology." made his assessment.
Amodei said that competition with countries such as China is the most difficult dilemma in taking steps on this issue, and argued that "by not selling chips to these countries, they will focus on increasing their own security and prevent the theft of their technologies."
Stating that a report submitted by Anthropic claimed that artificial intelligence may be used in biological terror attacks, Amodei said, "Even the Chinese do not want this. There is no room for biological terrorism in the interest of any government. Therefore, we need to talk to the Chinese and see if we can cooperate to restrict the speed of this technology." he said.
Amodei reiterated his call for cooperation to restrict this technology as much as possible, saying, "I think our hopes should be high and our expectations should be low." he said.
Asked about US President Donald Trump's meeting with Chinese President Xi Jinping, Amodei said, "The USA and China can agree that artificial intelligence should not be used to develop biological weapons or that models should not be presented that biological terrorists can use to develop biological weapons." he replied.
When asked about his responsibility for the course of artificial intelligence, Amodei replied that he found it "strange" that this technology was developed by a private company.
Stating that some people think that this technology should be developed by the government, Amodei said, "I agree with them. I am disturbed by this situation." He expressed that he felt responsible in this regard.
Amodei said that he "doesn't want to wake up one morning and find out that someone used Anthropic's artificial intelligence model to kill someone" and stated that everyone should play a role in this issue rather than a single person's responsibility.
Asked whether he would want to hand over this technology to the government, Amodei said: "I am concerned that a single government could misuse this technology as easily as a single company could. But I think that with democratically elected governments coming together, some form of control and shared governance could be given rather than handing over the company." he replied.
Amodei emphasized that there are serious dangers about artificial intelligence and said, "I think people have lied about the risks of this technology for a long time." he said.
Coxon's resignation
Jacob Coxon, who has been working as a researcher at OpenAI and Anthropic for the last three years, resigned from Anthropic, arguing that neither company acted responsibly enough regarding artificial intelligence.
Claiming that companies are in an uncontrolled race for self-improving artificial intelligence systems, Coxon said, "These will soon become superhuman systems that can hack anything, transform any area overnight, and obtain real power and resources." He used the following expressions.
Coxon also claimed that industry workers who developed artificial intelligence models seriously believed that these systems could "kill humanity" within 10 years.
