A Meta Platforms, in Mark Zuckerberg, announced last Tuesday (13) that it will provide access to the secret components of a new artificial intelligence model (AI) which promises to be “like a man”.
According to the company, this model, called I-JEPA, has the ability to analyze and complete unfinished images with greater precision than models already on the market.
see more
Thai food restaurant faces lawsuit after customer suffers…
Public tenders: Federal Government authorizes the opening of more than 3…
Unlike other generative AIs that only analyze nearby pixels, the I-JEPA model uses background knowledge about the world to fill in missing parts of the images, incorporating human-like reasoning into their creations.
This approach, championed by Meta's Lead Artificial Intelligence Scientist Yann LeCun, helps avoid common mistakes in AI-generated imagery, such as hands with extra fingers.
Meta, which owns social media such as Facebook, WhatsApp and Instagram, believes that sharing the models developed by its researchers can drive innovation in the artificial intelligence “market”.
In addition, innovations like the one presented now can help identify security gaps and reduce costs, in addition to propagating even more knowledge of the sector to the next generations.
Despite warnings from other companies in the industry about the potential dangers of the technology, executives at Meta refused to sign a statement backed by top executives from OpenAI, DeepMind, Microsoft and Google.
In a recently released open letter, tech moguls equated the risks of Artificial Intelligence to pandemics and wars.
Yann LeCun, considered one of the pioneers of Artificial Intelligence, advocated building security checks into AI systems rather than condemning the technology as a whole.
Additionally, Meta is incorporating generative Artificial Intelligence capabilities into its consumer products, such as ads capable of creating image backgrounds, and an Instagram product that can modify users' photos based on user prompts. text.