Algorithms promise to revamp entire industries, but raise concerns
09/06/2023
/i.s3.glbimg.com/v1/AUTH_37554604729d4b2f9f3eb9ad8a691345/internal_photos/bs/2023/R/8/6vbbw4ROSUllr3BMYJDA/mae9137.jpg)
Giovanni Cerri — Foto: Divulgação
At São Paulo-based Hospital das Clínicas (HC), Latin America’s largest hospital complex, artificial intelligence is helping to fight liver cancer, a silent tumor whose symptoms usually only come up when the disease is already at an advanced stage. An algorithm developed by MaChiron, an HC-backed startup, “reads” CT scans as they are taken. If it detects changes, it prioritizes the patient, greatly increasing their chances of treatment and cure. While it takes a physician an average of 40 to 50 minutes to interpret the scan, AI takes only 30 seconds.
“There is a huge potential for artificial intelligence in medicine. It is difficult to provide quality healthcare in the future without AI,” said professor Giovanni Cerri, president of InovaHC (innovation) and InRad (radiology) institutes of Hospital das Clínicas, affiliated with the School of Medicine of the University of São Paulo (USP). “[AI] is a tool that improves productivity, increases safety, and reduces the possibility of errors.”
Mr. Cerri’s words could be applied to most areas of economic activity. From risk analysis in financial services to harvest prediction sensors in agriculture, the applications of AI seem virtually limitless. Education, logistics, manufacturing, marketing, entertainment — to some degree, the technology is present in all of these industries, and its adoption is only likely to increase with the recent leap in generative AI, which is capable of producing text, photos, videos, and even other computer systems.
In a recent report, the consulting firm McKinsey calculates that generative AI alone can gross between $2.6 trillion and $4.4 trillion per year, based on 63 cases analyzed. In banking, the potential is between $200 billion and $340 billion, while in retail and consumer goods, the estimate is between $400 billion and $660 billion. Bloomberg Intelligence predicts that revenue from generative AI will grow to $1.3 trillion by 2032 from $40 billion last year. PwC predicts a $15.3 trillion contribution from AI overall by 2030.
In Brazil, spending on AI this year will exceed $1 billion for the first time, according to research firm IDC. This represents a 33% increase over last year. Investment in intelligent automation — as part of that amount — could exceed $214 million, up 17%.
But if artificial intelligence promises a new world of productivity — with several business opportunities and benefits for people — the challenges of mass adoption seem as great as the benefits.
The impact of algorithms will be inevitable in key areas such as employment, the environment, copyright, and digital security, economists and futurists warn. The challenge will be to harness the new capabilities while minimizing the risks.
Artificial intelligence is not new. Its history dates back to at least 70 years ago. Since then, it has faced periods of disinterest in academia and governments that have curtailed investment in research. There have been several so-called “AI winters” during this period.
A decade and a half ago, however, little-noticed factors outside of research began to combine to create the current, more explosive phase of artificial intelligence.
“If we had AI winters before, today we can say that we are in the middle of a tropical summer,” said Fabio Cozman, director of USP’s Center for Artificial Intelligence (C4AI) and professor at the university’s engineering school.
The interest in AI is driven by three main factors, Mr. Cozman said: the increased computing power of faster computers, the increase in the amount of data available, and advances in deep learning, which uses algorithms that work in a similar way to the human brain.
AI also owes its current popularity to recent movements in the field of generative artificial intelligence. It’s the brand new era of AI chatbots, such as ChatGPT, launched last November by the U.S. startup OpenAI, and Bard, launched this year by Google. Once limited to professionals and researchers, chatbots have put artificial intelligence in the hands of ordinary people.
“The computing power and the possibilities of implementing different types of AI have become more tangible. Also, the investment is not the same challenge it was years ago,” said Rodrigo Silva, a professor at the School of Computing and Informatics at Mackenzie Presbyterian University. “The scientific community believes this is AI’s finest hour, but that doesn’t mean we’re going to change the world overnight.”
AI research has migrated from the university to the corporate world, where most Ph.D.s are already concentrated and where most of the big language models, the heart of chatbots, come from. Global private investment in AI, at $91.9 billion last year, is down 26% from 2021. Still, it represented a value 18 times greater than the volume of investment in 2013.
In Brazil, consulting firm IDC expects AI spending to exceed $1 billion this year, a 33% increase from 2022. Just over 20% of large Brazilian companies say AI will take up more space in their budgets.
On the horizon of AI, employment is one of the biggest issues. According to the World Economic Forum, 69 million jobs will be created and 83 million will be eliminated by 2027. That’s a reduction of 14 million jobs — much of it due to the impact of artificial intelligence.
“AI is different from traditional automation,” said Mr. Cozman. “This time, the people who are going to lose their jobs are not necessarily the professionals with the lowest level of education. It’s middle-level people.” On the one hand, this makes it easier to relocate the professional, who would only have to go through training programs. However, creating policies to train workers is seen as an essential step in ensuring access to new jobs for the working population.
Lawmakers in several countries are considering passing legislation to regulate AI. Last year, 127 countries passed 37 laws containing the words “artificial intelligence,” according to a Stanford University report, compared to only one in 2016.
However, experts disagree on the level of regulation and whether it should exist at all. Countries such as Japan have adopted minimum parameters, while in Europe the regulatory framework is viewed with rigor.
In Brazil, the regulation is being analyzed by the Senate based on Bill No. 2,338/23 by Senator Rodrigo Pacheco.
One of the most debated issues is who will be responsible for overseeing the activities — a specific agency, an existing organization, or a group of entities. For Glauco Arbix, coordinator of USP’s Innovation and Competitiveness Observatory of the Institute of Advanced Studies, the complexity of the technology requires collegial oversight. “AI is a big data gobbler. But we are not trying to regulate data; we are trying to regulate a technology that has very large specific features and is beyond the control of any agency that exists today in Brazil.”
*Por João Luiz Rosa, Ivone Santana — São Paulo
Source: Valor International