laitimes

Giant Battle: AGI Comes to a Tipping Point?

author:51CTO

作者丨Mohit Pandey

Compilation丨Noah

出品 | 51CTO技术栈(微信号:blog51cto)

Once upon a time, artificial general intelligence seemed far away from us, but now things have changed. Will AGI be coming soon?

Some believe that this goal will be achieved next year, some predict that it will be achieved in 2029, and still others believe that it will never happen.

Singularity University founder and president of renowned computer scientist Ray Kurzweil recently spoke with Joe Rogan about his thoughts on artificial intelligence: "We're not quite there yet, but we're going to get there, and by 2029, AI will be able to match anyone." He added: "Actually, I'm considered a conservative. It is thought that this could happen next year or the year after. ”

Kurzweil recalls his prediction in 1999, when he predicted that artificial intelligence would be on par with humans by 2029. "For thirty years, people thought the idea was completely crazy," he added, "and in fact, Stanford University hosted a conference where hundreds of people from all over the world were invited to discuss my predictions." Although people think that day will eventually come, they do not think it will happen in 2029, but they think it will take 100 years. ”

Elon Musk retweeted on X and agreed with Kurzweil: "By 2029, AI will likely be smarter than all human intelligence combined." ”

Lex Fridman, an AI research scholar at the Massachusetts Institute of Technology, responded: "The next few years are going to be very interesting. I hope that in the end humanity will prevail. ”

In a tweet, Musk wrote: "Artificial intelligence could be smarter than any single human next year. By 2029, AI could be smarter than all human intelligence combined. ”

Kurzweil further explained that people often underestimate the speed of technological development. "In fact, technology doubles every 14 years," he says, adding that the common assumption is that technology is growing by only 2 percent a year.

Talking about the speed of computers, he said that the speed of computing has increased to 35 billion calculations per second, which is a 2,400 trillion times increase from 0.00007 calculations per second in 1939.

Last August, Saturnin Pugnnet, a founding member of Tools for Humanity, the company behind Worldcoin, initiated a discussion about AGI, claiming that "fusion technology and AGI are possible by 2030".

Pugnette's assertion that by the end of the decade, unlimited energy and intelligence will be possible, is reminiscent of the predictions made by OpenAI CEO Sam Altman a few years ago.

Altman has said that the cost of intelligence and energy is moving towards near-zero, and that by 2030, "the AI revolution and the combination of renewable energy and nuclear energy will allow us to achieve this".

The topic of whether AGI is coming has become one of the hottest topics online, with people from all walks of life voicing their opinions.

As the discussion continued, industry leaders expressed their views. Elon Musk has chosen an excellent time to file a lawsuit against OpenAI.

Just as OpenAI and other tech giants are investing in robotics company FigureAI, Musk has filed a lawsuit against OpenAI and Altman over the possibility that OpenAI may be close to achieving AGI. The lawsuit alleges that OpenAI has achieved AGI through GPT-4.

The same was true when OpenAI accidentally leaked Q* and was dubbed AGI by the masses. There is no doubt that OpenAI has been working on building AGI. Interestingly, the company has taken another step towards achieving this goal by tweaking the "core values" listed on its website to explicitly include a focus on AGI, which was not explicitly mentioned on the previous page.

Recently, Nvidia CEO Jensen Huang also shared his thoughts on AGI, saying that whether AGI can be achieved depends on how this goal is defined. "If I set up an AI with all the tests you can imagine, make a list of all the tests and present them in front of the computer science interface, I guess in five years, we're going to do well on every test," Huang said. ”

Huang also stressed that AGI may be far away because it's hard to achieve it as an engineer because engineers need clear goals. The path to AGI remains unclear as the assessment parameters are still evolving, the definition of AGI is constantly changing, and scientific and technical experts are divided on its goals.

Recently, however, Cognition AI launched an autonomous programming agent called Devin, which is capable of completing a variety of tasks. Not only that, but Devin can automatically generate code and write complete applications in a matter of minutes. Developers on the X platform are worried, afraid of losing their jobs, and are trying to upskill themselves to cope with the competition that comes with this major breakthrough.

Meta is also getting involved in the AGI space with the launch of the Llama 3 project. Meta CEO Mark Zuckerberg has also shown his determination on the path to pursuing artificial general intelligence. In his recent Instagram post, he declared: "Our long-term vision is to build universal intelligence and responsibly open source, ensuring that it is widely available and accessible to everyone." ”

One X platform user mentioned, "When we had the opportunity to name our AGI, we called it Devin. This seems to be a vision of what might happen in the future, such as maybe one day in 2029 (maybe sooner), AGI wakes up and introduces himself: "Hi, I'm Devin. ”

On the other hand, in a recent podcast interview with Friedman, Yann LeCun said, "You can't simply turn on a machine and suddenly have AGI, it's not something that happens overnight. This will be a gradual and non-mutational process. LeCun's point suggests that achieving true AGI is a long-term, step-by-step process, rather than an overnight goal.

Reference link: https://analyticsindiamag.com/agi-coming-soon/