@Unfiltered/AI: Bridging Bytes and Ballots | Edition 7
By Lacy Crawford, Alan Rosenblatt, PhD, and Craig Johnson, Unfiltered.Media
Welcome back to the Bridging Bytes and Ballots newsletter—helping you navigate the intersection of AI and politics. This week we are joined again by Unfiltered.Media partner Andrea (Haverdink) Priest and Craig Johnson. AI is advancing faster than I can type so...
Let’s get to it.
AI: A Possible Assistant or Possible Replacement
MIT News
Lacy: Artificial Intelligence in the hands of elites and the state is nightmare fuel for progressives, and I'd hope, for society at large. MIT economist Daron Acemoglu argues that AI can enhance human work rather than usurp it—a hopeful outlook that, to be clear, is not self-actualizing. This means it will take considerable effort from many to ensure that AI benefits, not harms, us all. Academics, advocates, and some politicians realize that humanity is on the cusp of AI that could either enhance our lives or lead to something more ominous. Andrea and Craig, I'll assume we agree that progress isn't automatic but requires intent. So, how can advocates, experts, and the general public help bend the arc to benefit everyone?
Andrea: Not to put on any rose-colored glasses, but I think we’ve already had front-row seats to the limitations on AI’s usefulness, and that should be the foundation on which we make this argument. At the start of this craze, managers and money-savers jumped to replace their email writers and other writers with ChatGPT… and were handed basic, contradictory, and nonfactual information as a result. Many of us have spent time on social media mocking AI’s inability to catch nuance or sarcasm, and other glaring failures. So at least with where the tech is right now, it truly does need to be an augmentation, rather than a replacement, of human work. On the flip side, however, Acemoglu’s point about it having taken “at least 100 years after the 18th-century onset of the Industrial Revolution… for the productivity gains of industrialization to be widely shared” is dead-on. Even if AI is used to augment, rather than replace, human work, it does not immediately mean workers get the benefits. They may even see more work hours or less pay, since bosses will argue that the machine is “doing the work for them.” Acemoglu is 100% right when he says that the only fix here will be increased worker power—and luckily that’s something that’s been ramping up rapidly in 2023 thanks to the hard work of labor unions across the country.
Craig: They can’t. The factors that have been unleashed already, much like the industrial revolution, assembly lines, or the transistor, were discoveries that led to seismic sea change in culture and society that none of the investors of these technologies could have imagined their impact and often the consequences of these technologies came too late for Government to prevent. These same factors could not be more true today. Congress is in no way equipped to regulate a technology as equally technology complex to them as a cell phone would be to the average 18th century worker. Advocates themselves are ill-equipped to handle these issues as well. They often have degrees in Political Science or Public Policy, with very few who understand both the technology and the political maneuvers required to regulate the tech. Even now there is an undercurrent on the left of avoiding anything that has AI in the name or that does work people don’t want to do as if this bottle can be uncorked. AI has been around us for decades, has been used for evil in every facet you can imagine, and will continue to be used for both good and bad in the future. The real concern remains not what we call “Artificial Intelligence” today but when someone does create something intelligent.
Artificial Advertising
Meta starts rolling out generative AI tools for all advertisers
Reuters
Lacy: Meta’s rollout of generative AI tools for advertisers appears to be a game changer, particularly for organizations and businesses with smaller budgets. Aside from questions of whether we can trust Meta, how do you see this being utilized to help candidates, organizations and businesses that may not have the resources to spend heavily on ads? And, what are your technical takes on what Llama 2 can do for ads?
Andrea: What I like about this announcement is that Meta calls the new tech a “testing playground.” Testing and learning from your data is the most important thing about social media advertising—and organic social media as well. This innovation will allow small businesses, candidates, and organizations to test the performance of their ads by creative and copy in real time, rather than having to wait on the creative to be delivered to them. That being said, if I were a betting person, I would put money down on the likelihood that most firms will still need a creative or team of creatives to provide them with content. At the end of the day, AI can’t truly replace the innovation and design skills of a seasoned and inventive graphic design professional—especially when that AI is only working off of what already exists.
Craig: I really like Andrea’s framing and I think it’s spot on. AI isn’t intelligent and is unlikely to create the perfect content for your advertising campaigns. What it is going to do is take the time to create 6 compelling ads to test from a 4-8 hour project to a 1-2 hour at most project. As digital professionals there are so many aspects of a program you have to pay attention to and oftentimes we can only execute imperfect programs. The logic being that if I can execute 3 programs at 80% but only one at 97% I should execute the three 80% programs. This allows you to take those 80% programs to 90%+ effectiveness without incurring more human hours spent on the project. AI is fundamentally an efficiency amplification machine for most white collar jobs.
AI and Design
Canva’s new AI tools automate boring, labor-intensive design tasks
The Verge
Lacy: So let’s get practical. Have AI products like Canva’s new Magic Studio or Adobe’s AI offering saved you all time and money in your day to day work? I’m excited to try these new tools and imagine that they will make life easier for graphic designers and teams that are under capacity and need a quick graphic. What I described is somewhat rudimentary. Have either of you used some of the more advanced offerings?
Andrea: The short answer is yes, 100%. The ability to remove a background with one click, generate an element you need, or to animate your design takes the hours spent on a single design or project down significantly. The most useful has actually been auto-subtitle tools, like Kapwing’s recent AI addition. But since I always add a caveat, I’ll make sure to say that these things do need to be perfected by hand. Those auto-subtitles are not going to be perfect, and they’re not going to look the way you want them to on every single frame. Same with Canva’s AI tools. And some of these tools are more helpful than others. For example, using AI to just auto-generate a video for you may cut down on time, but that doesn’t mean it’ll be an engaging video. You still need to be prepared to put in that work to optimize it for your platforms, your audience, and your message. And you may even decide to scrap the AI video entirely and just spend the time you need to make a truly great video. But what I’ve been saying about these tools overall is, don’t make things harder than they need to be. If there’s a product out there that cuts your captioning time in half, use it. Don’t make yourself hand caption every single video just because.
Craig: Generally I think this is good for the reasons Andrea said. That being said I have limited experience because I have an Andrea.