Meta Platforms (META) has released several new artificial intelligence (AI) models, including one capable of checking the work of other AIs for accuracy.

Meta's new AI model, called Self-Taught Evaluator, one of many models the company released on October 18, can improve the accuracy of responses from chatbots and other AI applications in complex fields such as science, Coding and math review problems.

Meta says it aims to address one of the most persistent criticisms of AI models like ChatGPT, namely that they often provide inaccurate or outdated information and answers to questions. Meta said it has developed a technique to break down complex problems and improve AI responses.

On the way to autonomous AI

Analysts say the ability to use artificial intelligence to evaluate AI is an important step toward ultimately developing autonomous AI applications and models that operate independently of human intervention. Ultimately, AI models will be able to learn from their own mistakes, self-correct and improve without the need for human intervention, researchers say.

While self-improving technology may sound like something out of a science fiction movie, analysts say it is part of the evolution toward “artificial general intelligence,” the point at which AI matches human intelligence and makes decisions on its own. Some researchers believe we will reach this point as early as 2030.

Other AI tools released by Meta on October 18 include an update to the company's image identification model and a tool that accelerates the generation of AI answers. META stock is up 64% this year.

Is META stock a buy?

Meta Platforms shares are unanimously rated “Strong Buy” by 47 Wall Street analysts. This rating is based on 42 buy, four hold, and one sell recommendations made over the past three months. The average META price target of $624.12 implies an upside of 7.91% from current levels.

Read more analyst reviews on META stock