The AI Daily Brief: Artificial Intelligence News and Analysis - An Optimistic Take on AI and Inequality

Episode Date: January 20, 2024

A reading and discussion based on "For all our fear of AI dystopia, it may help to level up society" https://www.ft.com/content/8f7b9b52-9243-4c34-af80-223522273ab4 ABOUT THE AI BREAKDOWN The AI Breakdown helps you understand the most important news and discussions in AI.  Subscribe to The AI Breakdown newsletter: https://theaibreakdown.beehiiv.com/subscribe Subscribe to The AI Breakdown on YouTube: https://www.youtube.com/@TheAIBreakdown Join the community: bit.ly/aibreakdown Learn more: http://breakdown.network/

Today on the AI Breakdown, we're talking about a very different vision for AI that's optimistic about how it might level the playing field. The AI breakdown is a daily podcast and video about the most important news and discussions in AI. Go to Breakdown.com for more information about our YouTube, our newsletter, and our Discord. Hello, friends. Welcome back to the AI breakdown. It being the weekend, you know that what we have here is a long read edition. Now with this one, we're going to read an essay that posits a particular view of AI, and then we're going to look at how the science and evidence reinforces that potential optimism. The piece is by Gillian Tet in the Financial Times and is called, for all our fear of AI dystopia, it may help to level up society. Gillian writes, Will I Am, the Grammy Award-winning rapper from the Black-Eyed Peas group, has long mesmerized millennials with his music.

This week, however, he grabbed the attention of economists, government ministers, and corporate leaders with a different tune. a vision for artificial intelligence. As debates about AI dominated this year's World Economic Forum meeting in Davos, Will I Am was among the loudest extolling the technology's putative power. That is partly because it is sparking his own creativity. This week he launched the first music radio show with a bot. However, there is another reason too. He thinks AI could pull marginalized people into the mainstream economy in future years and thus be a tool for social leveling. In particular, he told me, in a lively, expletive-laden speech on stage, he thinks AI will, quote, break down barriers for people who have nothing in a near unprecedented way.

Is this just another bit of Davos hype? Many might think so. It is true that in recent months a host of economists have predicted that AI will deliver a big boost to growth. Michael Spence, a Stanford University professor, for example, thinks it will add at least $4 trillion annually to global gross domestic product. But this chatter about a putative productivity miracle usually occurs amid fears about rising social inequalities due to the displacement of jobs. Indeed, at the start of this week's

WEF meeting, a survey from PWC revealed that a quarter of global chief executives expect generative AI to lead to headcount reductions of at least 5% this year. Meanwhile, the IMF predicted that AI will change 40% of all global jobs and 60% of those affected will be in developed countries. Editor's note, that's not exactly what that IMF piece said. It said that AI will change 40% of all global jobs, and in developed countries, that number would actually be 60%, slightly. different than how it was phrased here. Back to the essay. More alarming still, there is a widening digital divide in terms of uneven levels of digital literacy and access to the technology across platforms, and one that cannot be easily closed by education alone. No wonder a poll from Edelman

Public Relations Group shows that only 30% of the global public want to embrace AI, while 35% reject it. However, there are two key factors that help explain the alternative, more optimistic view about inclusion, espoused by Will I Am and others. One concerns how any of the AI might hit head, hand, and heart jobs, to cite British author David Goodhart, or those deploying skills. In the 20th century, digitization primarily hit jobs done by hand, and the displacement of factory workers in the West by robots did fuel income polarization, even if other jobs were created elsewhere, as economists such as David O'Tor have noted. But the difference between AI today and automation in the 20th century is that the new tech is hitting head jobs,

and to a lesser extent, heart rolls. As Josephine Teo, Singapore's digital minister, told a W.E.F. meeting. That hurts the elite professions arguably for the first time. Hence the squeals of alarm from pundits, which might leave some manual workers feeling some justified shot in Freuda observes Teo, herself a former union leader. The second factor is that history also shows that technological revolutions undermine incumbents, says Andrew McCaffee, an economist at MIT Business School. This is the case, be they companies, countries, or economic cohorts. That might seem hard to imagine today, since the elite who have developed and deployed AI have become fabulously wealthy. But if the acronym is presented in terms of augmented rather than

artificial intelligence, it is possible to see why hierarchies might yet be challenged by a tool that enables workers to execute complex cognitive tasks far more easily than before. Consider the jobs of writing legal contracts, advanced computer code, or medical diagnoses. Today, they are dominated by an educated elite. But if less educated workers can deploy AI to perform these roles in the future, that will break some of the barriers to entry for headwork. That is scary for the elite, not so much for the others. Hence why some AI leaders, such as James Menika of Alphabet, argued that this is already sparking a more positive attitude towards AI in the developing world than the developed world. And why social activists,

including Will I Am, hope that putting AI tools in the hands of more disadvantaged children will be empowering. The cynic in me would retort that there are endless obstacles that could torpedo this. Walthy elites are often extremely good at finding ways to protect their privilege and at building professional moats. And one grubby aspect of AI, is that its development to date has hitherto been dominated by elites in the West. This means that there is an urgent need to get wider participation in the creation of the technology. This is Alex Sato of Alliance for AI, a lobby group promoting access in African countries.

Without this, the tech will reinforce biases and hierarchies. Proactive, smart, and holistic government policies must be developed to bolster education and IT access and to ensure open source AI development. But here is the key point. If a rapper who grew up in a poor district in Los Angeles can dare to dream of a leveling upside for AI, other pundits should try to do so as well, even amid the dystopian chatter. I just wish the black-eyed peas would create a song that urges governments to deliver the policies to support this. It might finally grab attention from voters.

All right, so back to NLW here. I actually want to talk about that IMF survey that Gillian referenced, because I think that it shows just how unknown the real impact of AI is going to be. So when it comes to those 60% of jobs that they said would be affected in developed countries, the IMF said about half of those would see an augmentation and an increased productivity and potentially gains in their income. The other half could be significantly challenged or even replaced. I don't know if the numbers are exactly half and half, or if that's the effective equivalent of an IMF shrug, because who the hell knows? Now, what we are starting to be able to observe is that workers who are less experienced gain even more from generative AI than

their more experienced colleagues. For example, last summer we got a paper from MIT that found that contact center agents with access to a conversational assistant, overall saw a 14% boost in productivity. The largest gains came from the new and low-skilled workers. As an MIT blog post about this put it, in other words, the workers were upskilled, not replaced thanks to the technology. It said one of the paper's authors, Associate Professor Danielle Lee, generative AI seems to be able to decrease inequality in productivity, helping lower skilled workers significantly but with little effect on high-skilled workers. Without access to an AI tool, less experienced workers would slowly get better at their jobs. Now, they can get better faster.

Now, of course, this is just one study, but I think that broadly speaking, it does align with what many people have observed anecdotally. If you take the thing that you are best at and a thing that you barely know, and use AI to help you with both, the relative improvement around the thing that you barely know is going to dwarf the relative improvement of the thing you know really well, that creates, as MIT observed, a real flattening, which does come with consequences for the people at the top. This could lead to a flattening of wages as well, and thus there might be a backlash in a very different way than we've seen backlash as before. The flip side, though, and the optimistic take is that the increase of more skilled workers will simply increase the demand

for stuff that skilled workers can make. This is what people like Sam Altman believe, that the reason that AI isn't going to be just an absolute job's cataclysm is because humans don't have small wants and needs. We have an unlimited, voracious, never-ending, always ready to enlarge appetite for everything. The logic then goes that the market will adjust to the reality of these tools being available, and more specifically, the capacity of people using these tools in terms of what can be created. People then will want more of everything. To put it simply, the supply of more skill and more intelligence will lead to more demand for the things that more intelligence can make. I find myself largely in agreement with that position. At the same time, I don't think it's

quite a binary between that and a real chaotic, disruptive process along the way. I think that even if that is the ultimate outcome, and not necessarily on a super long time scale, there still will be enormous disruption along the way, and policymakers should be thinking about that. What I'm glad to see, though, is more of this sort of optimistic take making it into mainstream outlets. For the last few months, the only optimism around AI has come from dyed in the wool techno-optimists and accelerationsists like Mark Andreson. No disrespect to any of those voices, but they're not necessarily the voices that are going to capture attention among people that don't already largely agree with them. Now, I don't know if Will I Am is the greatest ambassador

either, but he at least represents something different than the Sand Hill Road set. Anyways, really interesting stuff. One of the more interesting things coming out of Davos over the last couple days. I hope you guys are having a great weekend wherever you are. Until next time, peace.