Openai Company introduced AI generation -GPT -5 New. Now he can answer the most difficult questions of the user, the company said. Experts doubt these statements and note that neuros networks still participate in predictions based on statistics.
In the presentation, GPT5 is called the “smart, fast and useful nerve network”. According to creative people, it will cause less ritual and try to please users more. The model finally learned how to say that I did not know, and did not invent the facts that didn't exist. In addition, the programming skills are seriously pumped, and now you can create a web application or game in the literal sense of a request, written to the first Deputy Director of the Duma Commission on Information Policy, Rotsit President Anton Gorelkin in his Telegram channel.
Today, users who often face a reliable conversation and a large amount of “water”, which the nerve network offers as an answer. It is not easy to separate particles from the Chaff: the problem is not even tons of useless information, but in the complexity of reality, sometimes it is easy to find reliable information about your verified resources.
And this is not surprising: today are high statistical prediction models, and not conscious systems, wills or independent thinking. GPT-5 is a powerful tool, but it still works on the principle: Give the best document based on probability, according to Alexey Tostigailo, a senior partner of the integrated IT ensine.
In order for anyone to become really “reasonable”, not all the GPT versions are updated. AI of the new generation should have architecture with abstract thinking ability, long -term memory and independence. In addition, different basic computers are necessary – quantum or nerves, as well as a new training model. It not only includes the absorption of text of the text, but also the interaction with the material and social world, as a person.
Vladimir Kravtsev, an expert in the field of AI and Axenix Analytics of Axenix said that so far, we improve the imitation, but still different from the thought
In addition to technical observation, it is important to take into account ethical issues: The creation of AI will require strict security frames so that the system can think, but do not act detrimental to one person. A person still needs to control and establish whose thought framework, and if so, then the thought becomes a very relative concept.
That is why we will see who the Viking's reasonable person only after appearing a new computing facility basically and the change in the machine model, an independent expert in the field of artificial intelligence, candidate for economic science Alexei Leron. Meanwhile, we are dealing with those who simulate more persuasive intelligence, which perfectly describes the process of thinking, but does not think.
However, the major breakthroughs in AI in the coming years are more than possible. The achievements of the development of the nerve network can become assistant Ji, who can solve complicated issues with minimum supervision, as well as systems equivalent to experts in an increasingly wider area. All this is a question for 5-10 years, Vasichev Vasilkov, CEO of Aimarketwave.