Wave of noise: Organizations
AI strategy for digital product organizations
“Generative Artificial Intelligence” promises to change the world unlike anything we have seen in our professional lives. And while very few people will in end remain unaffected by this development, as engineers, designers, and strategists, we find ourselves right in the thick of it. The tools built on top of services like GPT 4 are for us to bring to the market through our clients’ products and services in a way that hopefully improves and enriches the lives of those we build them for.
This article series outlines the approach we at DK&A are taking and the general, strategic position we have defined for ourselves and our clients.
A moment in time
Will I lose my job?
Will AI replace me?
Will I be manipulated or exploited?
“The Fourth Turning”, a book by William Strauss and Neil Howe published in 1996, presents a theory of generational seasonality. According to the book, 2023 places us right in the middle of a period called “Crisis”. “The Fourth Turning” tells us that this phase is characterized through intense social and political upheaval when the old order is destroyed, and a new one is born. And if we lift our heads for a moment and look around all things considered — climate change, the war in Ukraine, and the decline of democracy across western nations — “crisis” seems to be an apt description.
There’s much to be said about “The Fourth Turning”, and not all of it is positive, but it seems to describe our moment in time with eerie accuracy. On the flipside the 27-year-old book also makes for a hopeful storytelling device: The next phase, just around the corner, promises a new high — a period of cultural and economic prosperity when institutions are strong and society is stable, a time for the artists and the creators.
But for now, we are in a crisis, and looking at the feeds, it does quite feel that way, too. The world is asking: Will I lose my job? Will AI replace me? Will I be sued over the use of Midjourney? Will AI kill art? Will I be manipulated or exploited?
A sense of urgency
Impacts will be most visible in developed markets and white-collar occupations.
A wave of generative AI is rushing toward us with great speed, and whether we choose to or not, our industry is building the channels through which this wave will flow. Understanding this “wave” and figuring out how to ride it effectively, instead of being swept up by it, may prove to be a make-or-break factor for many digital product companies. It certainly appears that it will significantly rearrange the landscape once it hits.
Sam Altman, the CEO of OpenAI, explained in an interview a few months ago that the development of automation using AI had progressed differently than anticipated, even by them: Instead of automating and replacing physical labor, then cognitive labor, and only then (or maybe never) creativity, things went the opposite way.
Dall-E, Stable Diffusion, Midjourney, and even ChatGPT—generative AI went straight for what was considered a truly and uniquely human ability: to imagine and dream up. With that in mind, those of us who dream up and build digital products should be listening carefully.
We get paid to implement the output of creative processes using machines. Our ways of working have turned into a big metaphorical target on our backs. It’s highly likely that our industry will be completely transformed by what comes next.
What’s more, the effects will be widely felt in adjacent industries—our partners’ and customers’—as well. Goldman Sachs estimates in a recent report on the impact of generative AI on the global economy that impacts will be most visible in developed markets and white-collar occupations. That’s us.
Strategy without tactics is the slowest route to victory. Tactics without strategy is the noise before defeat.
Planning ahead
To ensure that we are still around when the dust settles, we will have to imagine a new ideal, future state, and build towards it in increments. This will require both long-term vision and short-term tactics.
Here’s how we at DK&A are preparing for tomorrow:
01. Tracking Change
The change represents the difference between the past and the present, a function of time. The rate of change in the context of an “AI revolution” might feel so high that society could struggle to withstand it without damage. However, it’s worth remembering that humanity has faced such situations before.
We need to monitor change to respond effectively and avoid succumbing to fear or hype. Tracking the rate of change will be crucial in understanding what’s to come and determining appropriate reactions.
Adoption Rate
Before GPT-4’s release, rumors about its parameter count were rampant and improbable. As parameter count proved hard to understand and a poor indicator of GPT model quality, the public focus shifted to the adoption rate.
Though comparing AI to services like Spotify or Instagram may not be useful, a carefully applied adoption rate could hint at AI’s growing usage. IBM’s 2022 AI adoption index states, “35% of companies reported using AI in their business, and an additional 42% reported they are exploring AI. AI adoption is growing steadily, up four points from 2021”.
Investment
Investment in AI research and development could be another metric. However, investment and hype are often linked, and the unstable post-pandemic world may make it unreliable for tracking change. Still, it reflects public and private interest in AI.
Despite cautiousness in traditional investments, the World Economic Forum expects continued momentum in AI investments, especially in industries directly impacted by economic and supply chain disruptions or mature industries able to scale AI adoption, such as financial services, retail, healthcare, and manufacturing.
Accessibility
“Accessibility” defines how many people can theoretically access AI tools daily for work or private lives, indicating a real-world impact. While many have used machine learning-enhanced systems, generative AI represents a distinct, new class.
To define accessibility for this class, we must consider digital literacy, general tool and infrastructure availability, cost, and regulatory support and public policy. Currently, this metric isn’t easily obtainable. To make matters more difficult — as legal and security concerns rise, companies may explore custom, on-premise solutions or limit access.
Regulators will likely address access, and while many initiatives focus on data privacy, governments worldwide will need to tackle this subject in the coming years.
02. Act Now but Act with Intent
As a digital product company, we’ve closely followed the development of underlying technology, implementing what made sense for our customers. As consultants, it’s generally crucial to stay ahead but be pragmatic and measured about doing so.
However, the current situation demands a different approach. Months ago, our leadership advised our entire staff to adopt at least one new AI tool in their daily toolchain, like ChatGPT, Stable Diffusion, Dall-e, Co-Pilot, or Luminous. This aimed to ensure a shared understanding of available tools, implementation, and industry direction.
This recommendation, echoed by others, wasn’t made without considering the risks: Microsoft and Github, as well as Stability AI and Midjourney, face copyright violation lawsuits. Using and recommending these tools to customers entails real risks, but we believe real-world experimentation is necessary.
If you know the enemy and know yourself, you need not fear the result of a hundred battles.
03. Learn, Adopt, Adjust
To understand the implications of generative AI for our businesses and products, a good first step might be to define their commonalities: ChatGPT spits out text, Midjourney and Stable Diffusion create images, and several companies are developing models that produce audio and video — these tools generate stuff. To be able to do that in a way that’s aligned with our expectations, generative AI models are all trained on existing datasets, such as websites, Wikipedia entries, books, photos, 3D renderings, and paintings.
For a given input, usually text, models then predict a “reasonable” output. So to oversimplify dramatically for the purpose at hand — these models provide us “more of the same”. This comes with pros and cons. On the one hand, “more of the same” means more code written, more apps delivered, and more software built. On the other hand, it leads to more unwanted content like spam, fake news, and division. But maybe most importantly, it also means that AI-generated content isn’t truly “new.”
Ted Chiang, author of “Arrival”, recently likened ChatGPT to a “Blurry JPEG of the web”. — Simplified, ChatGPT searches a compressed, encoded lexical space and generates sentences by chaining plausible results. It excels at sounding human but struggles with factfulness. ChatGPT doesn’t encode absolute truths or concepts of true and false but recognizes their relationship. It’s better suited for storytelling, but anything not found in the lexical space can only be approximated.
Research, Share, Experiment
Identifying the right use case for these tools will be paramount to success. While the models will become better, underlying issues — like the lack of fundamental human concepts like truth — will only be resolved through new types of models, integrating with external services like Wolfram Alpha for Math, for example. ChatGPT will soon roll out plugins that should address some of these issues as well as allow the model to access current information beyond its training cutoff date in 2021.
Once identified, we will have to validate, consider, and discuss new use cases to understand potential ethical, legal, or security concerns. Much of this discovery will be done through trial and error. This technology might be one of the first that will be discovered by us, rather than designed.
04. Team Sports
According to the report by Goldman Sachs, AI will impact 66% of all jobs over the course of a few years. To manage the impending transformation and ride the wave successfully, entire organizations will need to adopt new tools and ways of thinking, not just engineering or design.
We have to ask ourselves:
Which parts of our organization haven’t changed in a long time? How can AI serve HR, Finance, Leadership, etc.?
Which parts of our organization are easy to automate? Which tools and technologies are we currently using?
Which parts of our staff could be left behind due to their age, background, skillset, etc.?
And then act accordingly.
05. Learn to Learn
The most important strategy for weathering the current state of conflict and emerging as thriving, successful teams, companies, and societies is to establish and maybe in part rediscover our ability to learn.
Learning is part of DK&A, literally engraved in the company name. But even we will have to double down:
Culture
We need to nurture a culture of lifelong learning—not only at work but also in helping our employees stay curious, motivated, and engaged in the learning process wherever they are. This is where even many schools fall short today.
A study by the World Economic Forum (2018) revealed that employees with a strong learning culture are more engaged, productive, and adaptable to change. A mentality which enabled Google to develop impactful tools such as Gmail by granting employees 20% of their own time for personal development once upon a time.
Continuous learning enables our employees to develop new skills and adapt to the evolving job market. It makes them more versatile and resilient, allowing them to take on new roles and responsibilities as the need arises. Organizations remain more competitive and agile in the face of change.
Technology
We must embrace emerging technologies. As mentioned, we have prodded our staff to embrace new tools with some force regardless of existing concerns. Staying up-to-date on tools and technologies has always been important but will become a necessity.
Would have Netflix not embraced streaming over the shipping of DVDs the company would have very likely never become the market leader it is today. Embracing emerging technology allows companies to capitalize on new opportunities, streamline operations, and stay ahead of competitors. It also enables employees to develop valuable skills — especially when paired with a strong learning culture, ensuring their relevance in the evolving job market.
Experimentation
In that vein, we must also encourage experimentation and risk-taking, fostering a culture of innovation and creativity within any organization—not just in R&D, tech, and design. We must realize jointly that we are able and capable of scaling any challenges with an organized, flexible, creative approach.
A culture that supports experimentation allows employees to explore new ideas, learn from failures, and iterate on solutions. This encourages innovation and adaptability, which are crucial for success in the constantly changing job market.
Whether it’s the post-it note at 3M or the Echo at Amazon — companies who are willing to foster a culture that supports experimentation and the pushing of boundaries without the fear of failure often end up delivering new and highly innovative products before their competition.
Collaboration
Cross-functional collaboration fosters a diverse range of perspectives and skill sets, leading to more innovative and well-rounded solutions. This approach helps companies adapt to market changes by combining expertise from various areas, allowing them to identify new opportunities and tackle complex challenges.
To adjust more quickly to change, our staff must develop a broader understanding of the organization and the industry as a whole, both in terms of the current state and where things are heading.
Focus
Lastly, to be able to learn continuously with some ease, we must regain our ability to focus. We all know and feel it. Technology — the internet, social media, smartphones — has had a significant negative impact on our ability to concentrate.
And focus pays. — In his book “Deep Work” Cal Newport highlights the importance of focused, uninterrupted work for producing high-quality results. Companies like Basecamp have back in the days experimented with practices such as “no-talk Thursdays” where employees are encouraged to minimize distractions and focus on their work. We, humans, have taught ourselves that we are capable of multitasking. We have been led to believe that we can effectively juggle multiple tasks at once. While possible in some cases, research has shown that the human brain is simply not well-suited for multitasking. In fact, trying to do multiple things at once can actually reduce productivity and impair cognitive performance.
Whether we simply set clear boundaries, practice mindfulness, or enable more physical activity, we must regain our ability to focus our attention on the task at hand.
A case for hope
Any digital product organization will likely experience deeply transformative change during the fourth Industrial Revolution we are currently witnessing. However, with the right mindset, strategy, and a commitment to continuous learning, we can embrace these changes and thrive in the face of disruption.
By following the principles laid out, we can not only adapt to the challenges posed by generative AI but also capitalize on the opportunities it presents. However, it won’t stop there. We are likely on the cusp of a period of unceasing change. As we continue to explore the implications of AI in design and software development, and as the technology evolves, we must be prepared to adapt repeatedly and consistently. This will require maintaining a flexible and resilient mindset in order to thrive in the ever-changing landscape.
This is part 1 of a 3-part series on our strategic position regarding recent developments in Generative Artificial Intelligence. Read part 2 "Design" on our blog, the upcoming part 3 will discuss our take on key questions regarding Software Development, respectively.
