The #1 question for leaders in 2024: Will you be too early on AI, or too late?
Advances in AI audiovisual tools showcase the speed of change: See these 4 mind-blowing examples.
Welcome to 2024, which I suspect will be one of the top three most impactful years in the history of generative AI!1 I hope you had a great break over the holidays and are ready to buckle back up for a wild ride this year.
I came across a quote recently that really stuck with me:2
“There are only two ways to react to exponential change: too early or too late.”
So as you’re getting geared up for 2024, you should ask yourself: When it comes to generative AI, which do you want to be?
Here’s an example: 6 months ago I thought that AI audiovisual tools had limited immediate use. E.g., Midjourney, a best-in-class image generator, was good enough for some kinds of social media content, but required a lot of practice and could be frustrating to use.
But now everything is different: This category of tools has improved over the last 6 months more rapidly than any other tool category, and shows no signs of slowing down. Which means that the social change sector is already behind on thinking about how to leverage AI audiovisual tools.
Four impressive audiovisual AI tool releases
First, a few examples. Here are four tools or updates released recently that are blowing my mind a bit:
HeyGen video translate (video-to-video):
This tool actually came out about 5 months ago, but I haven’t shared it here yet and it’s still one of the most mindblowing use cases of AI for me so far. Basically, you upload a video of a person speaking in one language, like me explaining in English how HeyGen video translation works:
And you get an output video of the same person speaking whatever target language you chose, like me explaining in French (which I do not speak) how HeyGen video translation works.
Suno.ai (text-to-music-with-lyrics): In less than a minute’s worth of work, you can generate a minute’s worth of a brand-new songs, with lyrics, on any topic and in any style you want. The songs range from kinda-bad to honestly-surprisingly-decent. The UX isn’t ready for prime time — e.g., you can’t make any revisions to the clip once it’s been generated — but it’s very easy to see how a tool gets to the point where any teacher can make a catchy song reinforcing their lesson for the day, or anyone can create a pretty good soundtrack for a movie in a day or two.
Here’s a theme song Suno created for my writers group: https://app.suno.ai/song/44df02a6-8808-49e5-a38d-da3b9ff08907
Video Poet (text-to-video): I’ve been saying for the last 6 months that the text-to-video generators are tantalizing but not yet ready for prime-time, day-to-day use by most organizations. That’s still true, but there’ve been a few releases lately that are clear that we’re getting closer and closer. Most impressive is probably Video Poet, by Google. Google hasn’t released this tool for public use, so I have not tested it myself, but the examples they provide are pretty neat: https://sites.research.google/videopoet/.
Here’s a short film they made entirely out of Video Poet clips:
Midjourney v6 (text/image-to-image):
If you’re NOT already familiar with AI image generators, then look no further than this thread to understand the absolutely stunning progress that this technology has made in just 18 months: https://twitter.com/thatroblennon/status/1737977626344796169
If you ARE already familiar with/using AI image generators, you know that DALL-E 3’s release back in the Image Generation Dark Ages of Oct 2023 was a big leap from DALL-E 2. DALL-E 3 was the first time that anything was in practice more useful for me last year than Midjourney’s best model at the time.
Well, Midjourney has answered with their release of Midjourney v6. DALL-E 3 is still better at some things, but Midjourney has captured a lot of my usage share back with this release.
Here are some comparisons of Midjourney 6.0 against Midjourney 5.2 and DALL-E 3 (and remember, the leapfrogging cycles here are only about 3 months long):
PROMPT: Brown and purple intricate and fanciful logo for a chocolate company called "Wonka's". Include a nutrition label. In the upper left of the logo there should be a top hat.”
PROMPT: “Starbucks sign at a disco party”
Why do advances in AI audiovisual tools matter for your organization?
Now, I bet for many of you, your first thought is “DISINFORMATION 😱.” And you’re not wrong. We need to figure out as a society how to manage AI-generated deepfakes, and they will affect the work of many organizations.
But I’d encourage you to also think about the ways that all these tools can be used for good. Every organization with a communications department or social media presences should be thinking about revamping that work from the ground up. But from a program and direct impact perspective, there are also lots of enormous opportunities as well:
If you need to create narrative change in the general public, how can you reach vastly more people with vastly stickier content using these tools?
If you need to convey technical information (say, about policy or science) to non-technical audiences, how can these tools help you make more accessible information while also saving you money?
If you fundraise or manage other high-touch relationships, how can these tools help you create more compelling personalized communications — bringing your stakeholders along on your organization’s journey more effectively?
If you provide direct services to constituents (refugees? people with disabilities? people accessing government services?), how could these tools overhaul the way that they interact with your program and understand your offerings? We’re not far off from you being able to produce a compelling, heartwarming, funny video in any language with just a few hours work, about whatever information you need to convey to your constituents.
If you run trainings or work with educators, how can these tools help reach students/participants more effectively?
And the list could go on.
Ethical & legal concerns around AI audiovisual content production
Along with focusing on positive value and use cases for AI tools, I also do support organizations in managing risk related to AI usage, figuring out their AI policies should be, and sorting out their positions on ethical questions like the impact of AI content on artists.
There’s no doubt that AI audiovisual tools are going to significantly affect a lot of industries; they might cause serious financial hardship for some artists. Mission-driven organizations will have to balance the innovative potential of these tools against ethical and brand-risk considerations. This might involve grappling with issues around the creation of deepfakes, potential biases in AI, and maintaining transparency about the use of AI-generated content.
But despite these issues, there’s a fundamental reality at play: Whether you like it or not, AI AV tools are poised to transform consumption patterns, audience expectations, and business models across entire industries.
The question then is not just about whether to use these tools, but how to use them responsibly and effectively. I’d love to help you figure that out for your organization, and there may not be a one-size-fits-all approach. But complete avoidance is not an option.
Will your organization be too early on audiovisual tools, or too late?
Six months ago, it was arguably too early for most organizations to invest in audiovisual AI skills. But I predict that by the end of 2024, it will be too late. Not in the sense that you should no longer bother if you haven’t already, but in the sense that you will have missed large opportunities for impact. In some cases — e.g., for media companies — your delays may have jeopardized your own business model, as rivals pass you by or upstarts disrupt your industry.
If you’re interested in learning more about AI audiovisual tools, getting your team trained up on them, or developing an organizational strategy around them, please reach out to me:
And one final note: The fundamental dynamic of “too early or too late” isn’t just true of audiovisual tools, it’s true across the board for many applications of AI tools. For example, I think AI-powered research, knowledge management, and data analysis tools are on similar curves as AI audiovisual tools right now. Stay tuned for more on those areas of work!
With 2023 and 2022 being the other two years, of course :-)
Though he didn’t have a source for the quote, I came across it at Ethan Mollick’s Substack, which you should definitely subscribe to — he’s the best writer out there about the implications of AI for businesses.