Three Reasons for Hope in 2024
My second granddaughter is now 10 days old.
When I hold her, gazing down into the tiny face, I remember holding her older sister and, many years ago, holding their mother. And I think: Geeze, there are a lot of women in this house.
Just kidding. What I actually feel at these moments is a sense of hope. Call it irrational, but in that sweet little face I sense the promise of a better future.
Most of us have a source of hope in our lives. It could be our faith, our work, our loved ones – whatever it is, we surely need it, because for many of us, 2023 has been the latest in a succession of difficult years.
In this newsletter I'll be discussing three of the biggest, most complicated challenges we struggled with in 2023: Climate change, AI-driven fabrication, and political polarization. In each case, statistics helped identify the problem – and, in each case, statistics offer hope that 2024 will be a better year.
1. Climate change.
Frankly, you have to sift through a lot of bad news about climate change before you find grounds for hope.
2023 was the hottest year on record, greenhouse-gas emissions were the highest ever, record droughts were observed in some regions...the bad news seems endless, and we should be deeply concerned. Statistics were essential to identifying these problems (as I explain here), but in what sense could statistics – or anything else – be a source of hope?
(a) More compelling factoids.
For years, scientists have been warning us that climate change is not just a problem but a dire emergency. We've heard the message so often that numbness has set in. As one climate expert put it this week:
"We tack ‘mega’ on everything"....“It’s a megaheatwave, a megadrought, and a megastorm. And it just kind of loses its punch after a while."
All the same, the climate change statistics that made headlines in 2023 were especially striking. Think of this summer's heat waves, for instance. 44 days in a row of triple-digit weather in El Paso. 31 days above 110 in Phoenix. At one point, the ocean near Florida reached 101 degrees.
My hope is that these factoids are dramatic enough to spur more commitment to climate change solutions. At the moment, a key impediment to that happening continues to be political partisanship.
Pew Research Center surveys show that from 2009 to the present, the percentages of Democrats who consider climate change a "major threat" rose from around 60% to nearly 80%. During the same time period, the percentages of Republicans who feel that way shifted from 25% to – brace yourself – 23%.
If you're part of that 80%, one problem with this partisan divide is that it translates into dangerous action – or inaction – on the part of Republican leaders. For instance, it was almost exclusively states with Republican governors, including Florida, that turned down millions of dollars in federal funding this year to develop plans for reducing greenhouse gas emissions. Perhaps when the ocean near Florida reaches 110 degrees, the governor will be forced to reconsider. In the meantime, I'd encourage everyone to share and reflect on the scary statistics.
(b) Greater investment in renewables.
2023 was a record-setting year, and not just for weather. Global investment in renewable energy sources was by far the highest it has ever been. (In the U.S., we can thank the Inflation Reduction Act for many of the new or increased investments.)
Particularly important, at least in the short term, is the rapid implementation of solar power technologies. According to Business Insider, the U.S. as well as China and Europe each set records for the most solar installation in 2023, and solar power is now the cheapest form of electricity in many countries. I find the latter especially encouraging, although disparities still exist, with some of the poorest countries continuing to attract very little investment in any sort of renewable energy.
One more tidbit of hopeful news about renewables in the U.S. consists of changes to the federal rules for allowing new energy sources to connect to the grid. There's currently a huge backlog of projects waiting for approval that should begin to ease in 2024 and subsequent years, thanks to the new rules. (Here's a factoid that's even more striking than the summer weather statistics: The amount of renewable energy in projects waiting to connect to the grid (over 2,000 gigawatts) now exceeds the amount currently being produced.)
(c) More international consensus on the problem and potential solutions.
Pew data shows that international concern about climate change is high and continues to increase. At the UN Climate Change Conference (COP28), which ended this December 13, progress was made on several fronts. For instance:
–A "loss and damage" fund was created to compensate poor countries harmed by climate change. This fund will be administered by the World Bank, and the U.S. has promised to contribute $24.5 million dollars.
–Most participating countries signed an agreement to triple the capacity of renewable energy by 2030.
–All participants signed an agreement to transition away from fossil fuels "in a just, orderly and equitable manner" with the goal of net zero emissions by 2050.
There's a lot to pick on in the COP28 agreements. For one thing, they're not binding. $24.5 million dollars is a paltry sum given the scope of the problem (the Biden administration spent that much on relocating a single Native tribe threatened by climate change). Renewable energy capacity is increasing rapidly anyway. And, some folks are unhappy that the COP28 agreement fails to explicitly mention the phasing out of fossil fuels, as originally proposed. Still, incremental progress is better than no progress at all. This is the first COP agreement that openly calls for reduction in fossil fuels.
In sum, climate change has reached crisis proportions, but scary statistics, renewable energy investment, and international pledges offer hope that in 2024 we'll continue to see positive changes. (To learn more about what you can do, see here.)
2. AI-driven fabrication.
2023 was a breakout year for generative AI, but its successes have been marred by disruptive impacts including the loss of jobs, the spread of misinformation, the effects of racist algorithms, the enhancement of surveillance and military technologies, and, for some, the fear that robots will someday rule the world and find no use for us. (In earlier newsletters I discussed some of the ways statistics contributed to the development of the new AI programs.)
Here I'll be focusing on fabrication. Two types of great concern are deepfakes (AI-generated alterations that misrepresent what someone has done or said) and plagiarism (AI-generated content that someone misrepresents as their own work).
(a) Deepfakes.
We may chuckle now about the image of Pope Francis in a puffer, but it's not so amusing when people are scammed by audio deepfakes, or when an AI-generated forgery of an explosion near the Pentagon this May caused alarm and even a brief dip in the stock market.
Deepfakes are considered threats to everything from electoral security to personal privacy, but recent data also provide grounds for hope.
For the moment, the creation of persuasive deepfakes is resource-intensive, and people are still pretty good at identifying them (assuming they're looking closely). According to a study by MIT researchers currently in press, we're especially good at detecting AI-fabricated audiovisual presentations, as opposed to those consisting solely of audio or text. Simply put, the more information-rich the medium, the harder it is to fabricate content. Humans aren't perfect at spotting deepfakes – false positives and negatives do occur – but that's also the case for forged content created by conventional visual editing technologies.
Although you may be wondering what it means exactly that we're "pretty good" at calling out deepfakes, any statistics I cite would be misleading, because accuracy varies a lot depending on who's looking and what's being presented to them, and because AI technology is advancing so rapidly that accuracy data quickly becomes obsolete. We can say at least that in the MIT study, and in others, people tend to perform significantly better than chance.
Meanwhile, an arms race is underway between the technology used to create deepfakes and the technology used to detect them. Although there are debates about who's currently "winning", many experts believe that detection technology will continue to keep pace, particularly when multiple detection strategies are used.
This is not to say that deepfakes can be eradicated. They may always be with us. But I suspect they will turn out to be just one more technique of deception, rather than an unusually powerful one.
In his 2011 bestseller "The Code Book", Simon Singh shows that for thousands of years, code-makers and code-breakers have been locked in a predictable struggle: Once a type of code is broken, new codes are created that are then broken in turn. I suspect we'll see the same kind of cycle in the creation and detection of deepfake content, because both fabricators and detectives rely on the same technologies and skill sets. At the risk of sounding complacent, I find hope in the prospect of the detection strategies keeping pace, even though that's not as good as"winning".
(b) Plagiarism.
Some educators worry that generative AI programs like ChatGPT will cause (or have already caused) a spike in academic plagiarism. Why not let the chatbot do your homework or write that term paper for you?
As with deepfakes, I think the data tells us that AI will end up as merely one more weapon in the cheater's arsenal rather than creating a dramatic uptick in plagiarism.
For instance, since late October, education researchers at Stanford have been telling the media about a study in progress that shows no increase in cheating among high school students following the launch of ChatGPT.
Although the findings haven't been published yet, I find them both credible and a source of hope.
A long line of studies shows that the majority of students cheat, at least on occasion. In the Stanford study, for instance, both before and after ChatGPT's debut, 60 to 70% of high school students admitted to having cheated at least once in the previous month. This is fairly consistent with what prior studies have identified. And, as in all self-report cheating studies, the actual rates are probably higher, since not everyone cops to bad behavior, even on a supposedly anonymous survey.
In a February 2023 newsletter, I argued that we shouldn't expect AI to exacerbate the problem. Here's a revised version of what I wrote there:
With respect to cheating, generative AI programs create two potential concerns:
(1) An increase in the number of students who cheat at least once in their lives.
(2) An increase in the extent of academic dishonesty among students who already cheat.
Regarding the first concern, because most students cheat, at least on occasion, we may have a ceiling effect. There's not much room for AI bots, or anything else, to substantially increase the number of cheaters.
Presumably, some students won't cheat under any circumstances, owing to their moral beliefs and/or a fear of getting caught. The availability of new AI technologies won't affect the behavior of these students.
In short, if most students will cheat at least once by the time they graduate, and at least some of the ones who don't would never do so under any circumstances, it follows that new technologies won't create very many new cheaters.
Regarding the second concern, prior to the rollout of the new AI programs, students already had access to countless resources that support academic dishonesty, including Google, "homework help'' websites like Chegg, paper writing services, and other students. To the extent that AI simply replaces existing methods, there won't be a net increase in academic dishonesty. For example, when a student who doesn't want to write his own term paper uses ChatGPT instead of Chegg, ChatGPT doesn't increase the prevalence of cheating. Rather, one tool merely supplants another one, and the overall prevalence remains the same.
I'm not claiming this is good news. Overall rates of academic dishonesty are disturbing. I'm just saying there's some reassurance in the thought that AI may turn out to simply be another tool cheaters can use rather than a substantially more effective tool.
Effectiveness is the key issue, and here's where I find not just reassurance but hope: Educators are rapidly countering the threat of AI by adjusting their approaches to instruction and assessment, in some cases working with AI rather than against it. Meanwhile, plagiarism detection technologies continue to be refined and, in the long run, are expected to keep pace with the closely related technologies that support plagiarism in the first place.
Political polarization
What else can be said about a problem that we're so intimately and painfully familiar with? Scientific data, journalistic accounts, and personal observation all seem to paint the same dismal picture of increases in the extent of political polarization and its disruptive impacts.
Although this is a terrible problem, I want to suggest that it's not as terrible as we think.
(a) The loudest voices tend to be the most polarized (and polarizing).
There's plenty of evidence for this, so I'll just say it plainly: Polarized opinions tend to get a disproportionate amount of attention in the news and social media. The loudest voices don't speak for all of us.
(b) We misunderstand the extent of the partisan divide.
One of the best political analyses I read in 2023 is a white paper for the Carnegie Endowment for International Peace written by Rachel Kleinfeld. Among other things, Ms. Kleinfeld shows that most Americans underestimate the extent of shared policy beliefs among people from different political parties. In other words, we're less polarized than we realize.
Ms. Kleinfeld also notes that the extent of underestimation is greatest among people who are most politically active. So, you might say that the loudest voices are not only getting the most attention; they're also the least appreciative of the extent to which beliefs are shared across party lines (and thus their shouting distracts us from commonalities in those beliefs).
Ms. Kleinfeld also notes that even though our political beliefs are less polarized than we realize, there's a lot of affective polarization, which is a fancy way of saying that Democrats and Republicans often dislike each other. That's both bad news and a source of hope: If you can just get people to put aside their mutual dislike and talk to each other, they may find more common ground.
(c) Even researchers fall prey to caricature.
We tend to assume a lot of siloing in media preferences. You know, liberals only follow the New York Times, or CNN and MSNBC. Conservatives only follow the Wall Street Journal, or Fox and Breitbart. Many researchers share this assumption, because there's data that supports it.
However, these data come from self-report measures. In a study published this August, MIT researchers also tracked online behavior and found that most people engage with media representing a variety of ideological perspectives, even though on self-report surveys they claim exclusive interest in media consistent with their politics. Only a small group of people appear to be truly siloed.
I find hope in this too, because if you're exposed to other perspectives, there's always a chance you may be favorably influenced.
For instance, I spend some time on Foxnews.com every week. I don't think my liberal views have changed in response to anything I've read there, but this experience has at least gotten me more attuned to bias in liberal news outlets, including those that I respect way more than Fox.
Final thought
In the midst of writing this newsletter I had to pause and change a diaper, and, well, let's just say I was happy to get back to the writing. But I found a sort of metaphor in the mess. Climate change, AI fabrication, and political polarization might be described as three ways that we've crapped on ourselves. Underneath the crap the environment is still beautiful, the technology is still promising, and we still have a representative democracy that could, in theory, help create a better future.
Thanks for reading; I hope the new year starts off well for you!