AI Experts Say We’re on the Wrong Path to Achieving Human-Like AI

According to a group of hundreds of researchers in artificial intelligence, this field is currently following the wrongful general intelligence.

This insight was revealed in the Association for Progress at the 2025 Presidential Intelligence Committee (AAAI) on the future of artificial intelligence research. The prolonged report was developed by 24 artificial intelligence researchers whose experience ranges from the state of Amnesty International’s infrastructure to the social aspects of artificial intelligence.

The report included the main ready -made meals of each department, in addition to the community opinion department where the respondents were asked their ideas about the department.

The section on “AII Permanent Intelligence”, headed by the computer world at the Massachusetts Institute of Technology, Rodney Brooks, pointed out the description of the Gartner Hyge cycle, a five -phase course for technological noise. In November 2024, “Gartner estimated that the noise for artificial intelligence has just passed its climax and was down,” the report indicated. 79 % of the respondents in the community opinion department stated that the current general perceptions of artificial intelligence capabilities do not coincide with the reality of research and development artificial intelligence, as 90 % said that incompatibility impedes artificial intelligence research – 74 % of this number say that “artificial intelligence research trends are led by noise.”

“I have included the Gartner Hype course because it is something they used for years and it is a generalization on all noise followed by disappointment in many different fields,” Gizmodo told us in an email, so its presence and accuracy on many fields that tell us something about how to use caution in accepting the current levels of the component, at any time, at any time, at any time, at any time, at Any time, at any time. “

“I think large sections of public discourse on artificial intelligence are very acceptable to the level of noise as accurate,” Bruks added.

Artificial general intelligence (AGI) refers to intelligence at the human level: the virtual intelligence of a machine that explains information and learns from it as a human being. AGI is the sacred cup in this field, with repercussions on automation and efficiency across countless fields and disciplines. Consider any evil task that you do not want to spend much time doing it, from planning a trip to raising your taxes. AGI can be spread to reduce the burden of strong tasks, but also stimulates progress in other areas, from transportation to education and technology.

The most amazing majority – 76 % of 475 respondents – have simply witnessed the expansion of the current methods of artificial intelligence will not be enough to give AGI.

In general, responses refer to a cautious and moving approach forward: “Artificial intelligence researchers give priority to safety, moral governance, shareholder sharing, gradual innovation, and call for cooperative and responsible development rather than a race towards AGI,” the report was written.

Although noise distorted the research state – the current curricula that are raised for artificial intelligence do not place researchers on the optimal road towards AGI – technology has made leaps and limits.

“Five years ago, we could hardly conduct this conversation – artificial intelligence was limited to applications with which a high percentage of errors, such as the product recommendation, or where the field of knowledge was completely restricted, such as the classification of scientific images,” then, suddenly historically, General Amnesty International began working and reaching the audience’s attention through Chatbots such as Chatgpt. “

The realism of Amnesty International is “far from the solution”, reading the report, and I only answered the best LLMS on about half a set of questions correctly in the 2024 measurement test. But new training methods can improve the durability of these models, and new methods of organizing artificial intelligence can improve their performance.

“I think the next stage in improving the merit with confidence will be an alternative to individual artificial intelligence agents with teams of collaborative agents who are constantly verifying the facts and trying to keep each other honest.” “Most of the general public and scientific community – including the community of artificial intelligence – surpasses the quality of the best artificial intelligence systems today; artificial intelligence perception is about a year or two behind technology.”

Artificial intelligence does not go anywhere. After all, the Gartner Hype cycle does not end with “vanishing in forgetfulness”, but instead the “productivity plateau”. Various squares of artificial intelligence cases contain different levels of noise, but with all the noise about artificial intelligence – from the private sector, from government officials, from our families – the report is a refreshing reminder that artificial intelligence researchers think critically in their field. From the way artificial intelligence systems are built into the ways that are published in the world, there is room for innovation and improvement. Since we will not return to time without Amnesty International, the only trend is the front.