A report from the Stanford Human-Centered Artificial Intelligence Institute shows just how much the technology is in our lives, from its, at times, significant carbon footprint, to the vast amount of investment flooding into the industry, and even how different kinds of people feel about the powerful technology.
Called the AI Index, the report is assembled by an independent cross-discipline group of AI experts from academia and industry. Here are some of the key findings:
TECHNICAL PERFORMANCE
While 2022 was the year that generative AI broke into the public conversation and consciousness, the software still is prone to deliver falsehoods and in some cases nonsense with its trademark confidence, making it difficult to rely on despite its rapid adoption by companies like Microsoft into their core tools.
Despite those shortcomings, AI has proved to be a boon to science and has been used to "aid hydrogen fusion, improve the efficiency of matrix manipulation, and generate new antibodies" the report said.
The large amounts of computing power required to make the technology run, however, have raised concerns about its environmental effect. One AI program's training produced a carbon footprint 25 times that of a one-way plane trip from New York to San Francisco, according to one scientific study cited in the report.
ETHICAL ISSUES
The report pointed out that the content a software is trained on — inputs such as text and photos — can contribute to biased outputs. For example, "Text-to-image generators are routinely biased along gender dimensions, and chatbots like ChatGPT can be tricked into serving nefarious aims," the AI Index found.
One remedy could be training the AI on larger sets of data, with "new evidence (suggesting) that these issues can be somewhat mitigated after training larger models with instruction-tuning."
What is clear is that issues and controversies with the technology are on the rise, increasing by 26 times compared with 2012, according to tracking data from the AI, Algorithmic and Automation Incidents and Controversies database cited in the report.
ECONOMY
The U.S. is leading the way on investment in AI technology, and demand for AI skills increased from 2021 to 2022 in almost every sector of the American economy. "In 2022, the $47.4 billion invested in the U.S. was roughly 3.5 times the amount invested in the next highest country, China, at $13.4 billion" the report found.
And while the technology has the potential to eliminate up to 300 million jobs globally, one survey run by San Francisco code hosting site GitHub cited in the report found that most users of a text-to-code program called Copilot reported feeling more productive and working more quickly when using it, and that it increased their focus and work satisfaction.
EDUCATION AND DIVERSITY
AI has the potential to significantly alter how people learn, but the burgeoning industry has also turbocharged how many people are interested in working in the field. The report found that "The proportion of new Computer Science (CS) Ph.D graduates from U.S. universities who specialized in AI jumped to 19.1% in 2021, from 14.9% in 2020 and 10.2% in 2010."
But males continue to dominate those skilled ranks, with only 21.3% of new AI doctorates being female, a 3.2% increase from 2011.
More of those highly trained specialists are also choosing to pursue jobs in the industry instead of academia, especially as more and more private funding is poured into AI. The report found that more than 65% of people with a doctorate in AI took industry jobs in 2021, more than double the roughly 28% who took jobs in academia that same year.
FUNDING
Beyond private venture dollars, government spending is also spurring the industry forward. The Stanford report found that U.S. government spending related to AI has increased by an average of 52.5% since 2017.
Some of those are defense contracts, which have raised the ire of employees at companies like Google who objected to the technology potentially being used in military drones.
But a huge chunk of money still comes from private investors. Despite the hype being generated by the industry, global private investment in AI decreased more than 26% from 2021 to 2022, when $91.9 billion was poured into companies' from private coffers.
That was the first time in a decade that private dollars going to the industry decreased, but overall the trend has been up and to the right during that period, with private investment in AI 18 times larger in 2022 compared with 2013.
PUBLIC OPINION
Not everyone feels the same about the potential for AI to improve their lives. The Stanford report cited a 2022 IPSOS survey finding that men were more likely than women to say AI made their lives easier — 48% of men compared with 40% of women — trust companies to use the technology, and to feel that the benefits of AI outweigh the potential risks.
The survey found the most significant gap between men and women showed up when they were asked about their understanding of the technology — 65% of men said they understood it well compared with 53% of women.
In the U.S., those concerned about the effect the technology could have point to worries about automating away jobs for humans, privacy, and the fraying of human connections. But, the report found, those excited about the technology point to its potential to improve their lives by saving time and creating more efficiencies.
Another survey circulated to researchers in the AI field of natural language processing found that almost three-quarters of the respondents thought private AI companies had too much influence. Another "41% said that NLP (natural language processing) should be regulated, and 73% felt that AI could soon lead to revolutionary societal change."
©2023 the San Francisco Chronicle, Distributed by Tribune Content Agency, LLC.
