AI

AGI is suddenly a dinner table topic

First, let’s get the annoying actions of identifying the AGI from the road. In practice, it is a very foggy term and change that researchers or companies are constructed to build technology. But it usually indicates artificial intelligence in the future that outperforms humans in cognitive tasks. any Humans and any It makes the tasks that we are talking about all the teams in assessing the ability to achieve, safety and influence the labor markets, war and society. That is why the definition of AGI, despite its unfinished pursuit, is not enthusiastic but very important, as shown in a new paper published by Hugging Face and Google, among others. In the absence of this definition, my advice when you hear is that Agi is to ask yourself what Release From the mysterious term, the speaker means. (Don’t be afraid of the clarification request!)

Well, on the news. First, a new model of Amnesty International was launched from China called Manus last week. It is described by a promotional video of the model, which was designed to deal with “Agentic” tasks such as creating websites or conducting the analysis as “possible, a glimpse of AGI”. The model performs tasks in the real world on collective outsbox platforms such as Fiverr and UpWork, and the head of the Huging Face, called the AI ​​platform, “The most impressive artificial intelligence tool that I have ever tried.”

It is not clear how impressive is, but on this background – the idea of ​​AI AIS as a starting stone towards AGI – it was suitable for that New York Times Column writer Ezra Klein Bodkas on Tuesday to AGI. This also means that the concept may move quickly outside the artificial intelligence circles and in the world of dinner conversation. Klein joined Ben Bokanan, Professor George Town and a former special advisor to artificial intelligence at the White House.

They discussed a lot of things – what AGI means law enforcement and national security, and why the United States government finds it necessary to develop AGI before China – but the most controversial sectors have been about the potential impact of technology on labor markets. Klein said that if artificial intelligence is on the threshold of superiority in many cognitive tasks, it is better for lawmakers to circumvent what means the broadcasting of work on a large scale of human minds to algorithms for workers. The Democrats criticized the lack of a plan to a large extent.

We can consider this amplification of the balloon of fear, indicating that the AGI effect is imminent and attractive. After Gary Marcus, then he was lagging behind his proximity and firmness with a giant safety pin, is Gary Marcus, a professor of neural science at New York University and AGI critic who wrote to the points taken at the Klein Show.

Marcus notes that the recent news, including the amazing performance of the new Openai Chatgpt-4.5, indicates that AGI is more than three years away. He says basic technical problems are still existing despite contracts of research, and efforts to expand the scope of training and computing ability have reached the decrease in returns. The large, dominant language models may not be the thing that opens AGI. He says that the political field does not need more People who raise the alert about AGI, on the pretext that such a conversation benefits companies actually spending money to build it more than the public. Instead, we need more people who are wondering about allegations that AGI is imminent. However, Marcos does not suspect that AGI is possible. He only doubts the schedule.

After Marcus tried to empty it, the AGI balloon was detonated again. Three influential people – former CEO of Google, Eric Schmidt, CEO of Scent Ai aldr Wang, and director of the AI ​​Dan Hendrycks Center – published a paper entitled “SuperTelligence”.

Through “Superinteigence”, they mean Amnesty International that “it would decisively transcend the best individual experts in the world in almost every intellectual field,” Hendricks told me in an email. “The most safety knowledge tasks are penetration, virus science, research and independent development-where human experience can lead to severe risks.”

2025-03-11 09:00:00

Related Articles

Back to top button