Artificial intelligence (AI) is a growing field that makes machines smarter, faster and more intelligent every year. However, it’s still not up to par with the hype created by large technology companies around the globe.

AI may excel in chess play, but AI struggles with more complex tasks. An example of this is a 7-year-old who has more intelligence than all the AI systems.

Edward Grefenstette is a researcher at MIT. He said that AI algorithms can be used to tackle individual tasks or those with a low degree of variability. MetaCNBC was told by AI, previously Facebook AI Research.

He said, “But, the reality of change is significant, which is something we’re bad at understanding in our training algorithms. This leads to brittle intelligence.”

Grefenstette stated that AI experts have shown there is a way to adapt AI training methods for changing tasks or environments, which results in stronger agents. Grefenstette believes that this year will see more scientific and industrial applications. This will result in “noticeable leaps.”

Even though AI is still a ways off from human-level intelligence, there are many people who have already made progress. GoogleFacebook (Meta), and AmazonHuman resources have been invested in AI researchers capable of improving everything, from voice assistants and search engines to hiring billions of dollar-earning AI specialists. aspects of the so-called “metaverse.”

CNBC spoke with Beth Singler from the University of Cambridge who is an anthropologist and studies AI and robotics. Singler said that the claims regarding the efficacy and realness of AI in areas that are currently being called the metaverse in 2022 will be more widespread as more money and awareness of the concept of the “metaverse,” as well as the fact that more people start to use the term.

Singler warned, however, that in 2022 there might not be enough discussion about the effects of the metaverse upon people’s identities, rights, and communities.

Gary Marcus was an AI scientist, who helped to sell the company. UberCNBC has spoken with and, who is also the current executive chairman for Robust AI. CNBC believes that 2022’s most significant AI breakthrough will be one the rest of the world won’t see immediately.

According to him, “The process from laboratory discovery to practice can take many years.” He also said that deep learning still has a lot of work to do. Deep learning, an AI area that attempts mimicking the activity of layers of neurons in the human brain in order to recognise complex patterns in data, is called deep learning.

Marcus says that the biggest challenge in AI today is finding a “good way to combine all the world’s vast science-technology knowledge” using deep learning. Deep learning cannot leverage all this knowledge, and is instead stuck trying to figure everything out from scratch,” Marcus said.

Marcus stated, “I expect there will be progress in this area this year, that will eventually be transformative, toward what I called hybrid system, but it’ll take another few years before you see significant dividends.” The first drug in which AI was a significant part of the discovery process is what we will likely see next year.

AI software has been developed by the company that accurately measures distances. predict the structure that proteins will foldIn just days, we solved a 50 year-old “grand problem” that could lead to better drug discovery and understanding.

CNBC interviewed Neil Lawrence from the University of Cambridge about machine learning. He said that DeepMind would like to focus more on big science problems in 2022.

Language models — AI systems that can generate convincing text, converse with humans, respond to questions, and more — are also set to improve in 2022.

This is the most well-known model of language. OpenAI’s GPT-3DeepMind said in DecemberThe new “RETRO” model of language can surpass other models 25 times its own size.

Catherine Breslin (a machine-learning scientist, who worked for Amazon Alexa), believes Big Tech will move toward bigger and more complex language models in the coming year.

CNBC was informed by Breslin who is now the CEO of Kingfisher Labs an AI consulting firm. He said that models will be developed that can combine speech, vision and language capabilities, as opposed to treating them separately.

Nathan Benaich (Venture Capitalist, Air Street Capital) and co-author of this annual State of AI reportCNBC’s, reported that new companies are likely to use language models for predicting the most efficient RNA (ribonucleic acids) sequences.

He said, “Last Year we saw the impact of RNA technology as novel covid vaccinations, many built on this technology brought an end nation-wide lockdowns.” “This year we’ll see new AI-first RNA therapeutic businesses. These new companies will use language models to determine the best RNA sequences for a particular disease to speed up discovery of new medicines and vaccines.