The future of generative aI: early challenges for IT, and opportunities
I recently attended the annual scenario planning session for friend and colleague Rob Salkowitz’s ‘Future of Marketing’ class in the Communications Leadership program at the University of Washington. We introduced his students to scenario planning and asked them to think about how the technologies of the moment may play out against different social, technological, economic, environmental, and political backdrops over the next decade.
The scenarios simultaneously suggest that under the right circumstances, our fears for an overtly AI-fueled world where data and algorithms rule may manifest—or that AI’s history of over-promising may just as likely lead us to a fourth AI winter—or equally lead to a dysfunctional future where AI-driven misinformation forces those who want to maintain civilization into creating islands of “truth.”
As I assert to my learners, there is no data from the future. At this point, all of these futures are equally valid—our choices create the future. Scenario planning helps us make better-informed decisions by helping people experience the implications of those decisions through narratives that describe how choices manifest and interact.
Few IT organizations explore technologies through the lens of scenario planning. Hopefully, these ten takeaways from our class’s scenarios will encourage readers to explore our scenarios (which can be found here) so that they can employ the technique to create more robust IT strategies.
What IT leaders can learn about the Future of Generative AI from scenario planning
When a new technology becomes the center of discussion, it disrupts multiple assumptions underlying technology architectures and business solutions. Planners often dismiss the technology as a fad, panic, and over-focus on its threat to beliefs or wantonly adopt it with little thought about its implications.
Scenario planning offers a middle road, one where multiple narratives about the future, populated with uncertainties of varying values depending on the scenario logic, expose the weaknesses in ideas and technologies and the interconnections between technology, the economy, society, politics and the planet that may fuel change or curtail it.
Scenario planning builds a set of lenses that offer different perspectives about the future. Those lenses can help teams avoid surprises, drive innovation, or better navigate change by pre-considering which triggers to monitor and which contingencies to deploy when shifts occur.
Ten Take-Aways about the Future of Generative AI from Future of Marketing Scenarios
- We can’t assume the current movement toward generative AI ubiquity will persist. Social and political factors may disrupt adoption through legislative action or social pushback.
- It is too early to determine the limits of generative AI, but more than one scenario suggests severe tensions between creativity and mediocrity may challenge the long-term viability of the technology.
- The best uses of AI will focus on solving real problems. In the near term, the proliferation of generative AI may prove more of a distraction than a productive partner. It will take time for consistent value to emerge.
- There will be many implementations of generative AI, and they will not share equal veracity for the truth; some may be trained to support a bias or purposefully sow disinformation. Discussions of AI ethics are misplaced. Humans must deploy ethics in their use of technology; it cannot be embedded in it.
- Many tools will adopt generative AI in the near term with varying degrees of success in offering value. Evaluate AI features as you would other new features, including what people need to know to use them effectively.
- AI will likely create new challenges and problems, such as the proliferation of text-based content of questionable value.
- AI is just as likely to become domineering as it is collaborative, depending on human choices.
- AI will become a critical factor in analyzing an organization’s data, creating an opportunity to deeply understand organizations in the same way we use data to study and model biology or the cosmos.
- AI will displace some work. As a result, organizations must determine how they balance their objectives against innovation or efficiency by deciding which skills and how many people they require to be in the organization that meets their vision.
- IT needs to understand the current modes of AI to successfully apply them, when to reject them, and when to experiment with them. In addition, the new AI will create unknown risks often glossed over in vendor marketing materials that tout opportunities.
What studying history tells us about the future of generative AI
Most scenario planners adopt the maxim, “Look back twice as far as you look ahead.” Looking back to the early 2000s when I was still an analyst with Forrester Research, I attended a Microsoft Office briefing in Redmond, WA. Microsoft VP Kurt DelBene shared the prospects for a panel in Office documents that would suggest related content, consolidate communications around the documents, and reveal relevant content from other sources that could be used to inform, substantiate, or refute points in a document.
While AI now offers a panel in tools like PowerPoint, it formats rather than informs. DelBene’s vision of AI in Office remains a vision twenty years later. Even with the promise of integrating generative AI into Microsoft Office, core computing technology, from Microsoft Windows to MacOS and Chrome, to Azure and Amazon Cloud, remain ignorant to the knowledge that could be discovered on local drives and in private and corporate cloud repositories.
Generative AI seeks to offer universal access to the wealth of information on the Internet, seemingly homogenized and reduced to its most essential, least offensive, and most mundane representation. But, as Miro recently demonstrated, that isn’t a bad thing if all you want is a list of ideas for how to help a team communicate pushed into a mind map or to autogenerate sticky notes to start a brainstorming session.
Innovation, however, happens in scribbles and sketches in notebooks, conversational exchanges through e-mails and community feeds, half-finished presentations, and mind maps full of gaps—because innovation occurs at the edge of knowledge where people create new patterns.
The AI of this moment, like DALL·E 2 and ChatGPT, offer new tools that will inevitability face limits to their underlying model. They may turn out to be a fad, fading away like frosted tips, Four Square mayorships, or Netbooks—or evolve into something different—and at this point, I can only say different—not better or worse. GPT may prove as fundamental as the spreadsheet or as neglected as Microsoft’s SmartArt. We need scenarios to help us peek around the corners of the future. So far, the narratives go either way.
During the expert system era, AI tools like VP-Expert were touted as universal solutions accessible to the masses—anyone could develop an expert system. However, as the expert system version of AI Winter sent its chills through the market, pundits realized just how many copies of VP-Expert were tucked into file cabinets, never opened, or sitting in seldom accessed DOS directories.
Storage and processing are cheap. That ChatGPT even exists proves that point. But the wealth of processing and storage does not imply causality between tools that employ them and the value those tools provide. I’m not advocating for a new AI Winter, but I won’t be surprised when the clouds form.
For more serious insights on strategy, click here.