An Ubtech Walker X Robot plays chess during the 2021 World Artificial Intelligence Conference (WAIC) at Shanghai World Expo Center on July 8, 2021 in Shanghai, China.
VCG | VCG via Getty Images
Machines are getting smarter every year, but artificial intelligence has yet to live up to the hype generated by some of the world’s biggest tech companies.
AI can excel at specific narrow tasks like playing chess but it struggles to do many things well. For example, a seven-year-old child has a much broader intelligence than any current AI system.
Edward Grefenstette, a research scientist at Meta AI, formerly Facebook AI Research, told CNBC.
“However, the real world includes the potential for significant change, a dynamic that we are very poor at capturing in our training algorithms, which produces brittle intelligence,” he added.
AI researchers have begun to show that there are ways to effectively adapt AI training methods to changing environments or tasks, resulting in more powerful actors, Grefenstette said. He believes there will be more industrial and scientific applications of such methods this year that will make “remarkable leaps and bounds.”
While AI still has a long way to go before achieving anything like human-level intelligence, it hasn’t stopped it. Google, Facebook (Meta) and Amazon invest billions of dollars in hiring talented AI researchers who are capable of improving everything from search engines and voice assistants to aspects of the so-called “metaverse”.
Anthropologist Beth Singler, who studies AI and robotics at the University of Cambridge, told CNBC that claims about the effectiveness and practicality of AI in spaces currently labeled as the metaverse will become widespread. in 2022 as more money is invested in the sector and the public begins to recognize the “metaverse” as both a term and a concept.
Singler also warned that there could be “too little discussion” in 2022 about the impact of the metaverse on people’s “identity, community and rights”.
Gary Marcus, a scientist who sold an AI startup to Uber and now executive chairman of another company called Robust AI, told CNBC that the most important AI breakthrough in 2022 will likely be one that the world doesn’t immediately see.
The cycle from lab discovery to reality can take years, he said, adding that the field of deep learning still has a long way to go. Deep learning is an area of AI that tries to mimic the behavior of layers of neurons in the brain to learn to recognize complex patterns in data.
Marcus believes that the most important challenge facing AI right now is “finding a good way to combine all of the world’s vast scientific and technological knowledge” with deep learning. Currently, “deep learning can’t take advantage of all that knowledge and is instead stuck trying to relearn everything from scratch,” he said.
Marcus added: “I predict there will be progress this year, which will eventually be transformative, towards what I call a hybrid system, but it will be a few years before we can see it. see great profits”. “What we’re likely to see this year or next is the first drug where AI plays an important role in the discovery process.”
One of the biggest AI breakthroughs of the past few years has come from the London-based DeepMind research lab, which is owned by Alphabet.
The company has successfully created AI software that can predict the structure in which the protein will fold In just a few days, solving a 50-year-old “major challenge” could pave the way for a better understanding of disease and drug discovery.
Neil Lawrence, a professor of machine learning at the University of Cambridge, told CNBC that he hopes DeepMind will target more of the larger scientific questions by 2022.
Language models — AI systems that can generate persuasive text, converse with humans, answer questions, and more — are also set to improve by 2022.
Catherine Breslin, a machine learning scientist who worked on Amazon Alexa, thinks Big Tech will race towards larger and larger language models next year.
Breslin, who now runs AI consulting firm Kingfisher Labs, told CNBC that there will also be a move toward models that combine vision, speech and language abilities, rather than treating them as tasks. separate service.
Nathan Benaich, a venture capitalist with Air Street Capital and co-author of the annual magazine AI status report, told CNBC that a host of new companies will likely use language models to predict the most efficient RNA (ribonucleic acid) sequence.
“Last year we saw the impact of RNA technologies as a new covid vaccine, many of them built on top of this, ending the deadlock,” he said. nationwide rule”. “This year, I believe we’ll see a new crop of AI-first RNA therapy companies. Using language modeling to predict the most efficient RNA sequences targets a diseases of interest, these new companies can dramatically speed up the time it takes to discover new drugs and vaccines.”
While some progress may still be there, there are still major concerns around the ethics of AI, which can be highly discriminatory and biased when trained on certain sets. certain data. AI systems are also being used to power automatic weapons and create fake pornography.
Verena Rieser, professor of conversational AI at Heriot-Watt University in Edinburgh, told CNBC that there will be a stronger focus on the ethical questions surrounding AI by 2022.
“I don’t know if AI can do a lot of ‘new’ things by the end of 2022 but hopefully it will do it better,” she said, adding that this means it will be fairer, less more biased and more comprehensive.
Samim Winiger, an independent AI researcher who used to work for a Big Tech company, added that he believes there will be revelations surrounding the use of machine learning models in financial markets, espionage and health care.
“It will raise big questions about privacy, legality, ethics, and economics,” he told CNBC.