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

According to a panel of hundreds of artificial intelligence researchers, the field is currently following the artificial ordinary detective in the wrong way.

The President’s Association for Advancement of Artificial Intelligence (AAI) was published in the Panel about the future of 2025 AI research. The Long report Twenty -two AI researchers kept together with the Social aspects of artificial intelligence from the state of AI to the state.

The report includes a major adoption for each category as well as a community opinion section where their own thoughts were asked about the respondents’ category.

The section of “AI Persia vs Reality”, chaired by MIT computer scientist Rodney Brooks, mentions the Gartner Hype Chakra, which is a simple five-phase cycle for technology hype. In November 2021, Gartner “assumed that the generator AI’s hype had just reached the top and was in Downswing,” the report said. 795% of the Community Opinion Department respondents say that the current public ideas about the capacity of AI do not match the reality of AI research and development, 90% says that Amil is obstructing AI research – 74% of this number says “AI studies are driven by hype.”

Artificial General Wisdom (AGI) Human-level intellect refers to: a machine’s estimate that explains information and learn as a human being from it. AGI is a sacred grail in the field, countless fields and branches for automation and skills. Consider that you do not want to spend much time on any of the Metel’s Task to start traveling to file your tax. The AGI can be deployed to reduce the burden of rotation, but from transport to education and technology can make progress in other areas.

Wonder majority among 475 respondents – 76% – Side that is not enough to produce AGIs of the current approach to AI simply.

“Overall, the reactions are still a warned-movie procedure to indicate: AI researchers prioritize protection, moral administration, benefits and gradual innovation, advise in favor of the AGI instead of the AGI,” wrote in the report.

Despite the distortion of the study of hype research – and the current method in AI is not in the most favorable way to the researchers of the AGI – the technology has jumped and created the boundaries.

“Five years ago, we could rarely do this conversation – AI was limited to applications where high percent of the defects could be tolerated, such as the recommendation of the product, or where the domain of knowledge was strictly canceled, such as scientific images was classified,” Henry Kautz, Virginia University, Virginia University, and the University of Virginia. “Then, suddenly, the Historic Tihasik language, the General AI started working and the public began to notice the public through a chatbut like ChatzPT.”

The AI ​​factual is “far from the solution”, the report is written and the best LLMS has only answered about half of a set of questions correctly in the 2024 benchmark exam. However, new training methods can improve the view of those models and new ways to organize AI can make their performance even better.

“I believe that the next stage of credibility improvement is to be replaced with the cooperative parties of agents of individual AI agents that constantly tried to keep others truth-cheeks and try to keep each other honest,” added Coutus. “The scientific community, along with the community of AI researchers, is under the quality of the best AI system today; the concept of AI is about one or two years behind the concept of AI.”

AI is not going anywhere; After all, the Gartner’s hype cycle does not end with “faded with forgotten”, instead “the plateau of productivity”. There are different areas of the use of AI, but from the private sector to government officials, Hake, our own families have all the noise about AI – this report is a refreshing reminder that AI researchers are thinking very critically about their field status. There are opportunities for innovation and improvement from the way they are deployed in the world. Since we are not going back at any time without AI, the only side is ahead.