Three ways AI will reshape the digital workplace (and two ways it won’t)
By now, you have read plenty of hype about the capabilities of generative AI (GenAI) in the digital workplace — with terms like “disruptive” and “evolutionary” and “game changing” thrown about. I’m not here to disagree — all of those might be true, depending on how you choose to move forward.
AI can do many things that were considered farfetched just a few years ago, but I’m more interested in exploring how GenAI should and should not be used within a modern enterprise. It’s incredibly powerful, but must be guided by a set of social values and human-centric principles.
That being said, let’s look at a few of the top ways that AI will impact the digital workplace, and how it can be governed.
The top AI impacts on the digital workplace
1. AI will transform how we manage and consume knowledge.
According to the familiar cliché, “data is the new oil,” but this is only partially correct. If it means that data trapped underground is of no use until it is brought to the surface and refined, then yes. At that point it becomes knowledge — which is what today’s businesses truly run on.
The appeal of GenAI is its ability to refine large amounts of data and make that knowledge accessible to everyone. More than that, it has the ability to speak to every individual in a context that they can understand. In other words, it can use the same pool of information to generate knowledge that is consumed by different people for to meet their individual needs. This is at the core of leveraging technology to create a human-centric employee experience.
If you need to understand your company’s new product from a sales or marketing perspective, AI can tell you its strengths and weaknesses as compared to a competitor. If you ask it a technical question, it can give you a feature comparison or provide step-by-step installation instructions — all from the same content.
Rather than creating separate documentation for each user persona, a digital workplace with an AI-driven knowledge repository can provide personalized, interactive access to information that is tailored to each employee’s cognitive and behavioral needs.
The AI manages the mundane task of sorting through endless volumes of content, freeing your employees to focus on what matters — adding the human touch by applying their creativity and intuition.
2. It will help us prepare and perform better in our tasks.
As I explained above, the biggest strength of AI is that it can interpret huge amounts of data and provide information that is relevant to us as individuals. When it comes to quickly and accurately sorting and retrieving information, AI is simply unmatched.
However, it can be leveraged for more than just dispensing answers. One of the best uses is to provide individualized learning and training.
AI can understand a person’s skill or knowledge level in a certain topic, then personalize the training material to suit their learning style and individual goals. It’s not just capable of answering your questions, but can flip the tables. AI can play different roles for experiential learners or simulate different personalities to add a social and emotional component to the learning.
This is a remarkably powerful capability, but you have to be an active participant. Employees need to embrace the change and learn how to structure prompts and engage with AI to take full advantage.
They also need to share a certain amount of personal information for it to deliver the most value in terms of personalization. Hence, it must operate within strong ethical guardrails. More on that below.
3. It will enable us to create and collaborate in new ways.
AI can help us collaboratively analyze data, brainstorm ideas and create new content, sales pitches or presentations, but the sophistication (and value) of the output depends on a few key factors.
Today’s generative AI models rely on a variety of source data, inputs and preferences, and the output depends on the amount and quality of the source information. For AI to really work, you cannot indiscriminately dump all information into a repository and let the AI work its magic. Your document management function must be aligned with your business objectives, ensuring that documents fulfill a strategic function.
It also requires a proactive approach to regularly update these documents with the latest internal and competitive information. If not, you run the risk of multiplying errors by drawing conclusions from irrelevant or incorrect source data.
Finally, and most importantly, the knowledge and information that AI unlocks must be accessible to all. AI itself is inherently democratic, enhancing accessibility and enabling knowledge creators and consumers to share information in your digital workspace.
However, to effectively use AI for collaboration, your enterprise must design with purpose — creating a flexible, human-centric ecosystem that encourages participation and inclusion.
AI can bring major benefits to the digital workplace, but for now it can only facilitate and enhance our interactions. We humans still have to take ownership for the decisions we make.
The top challenges of the AI-powered workplace
Now that we have defined some of the potential of generative AI, let’s look at some of its limitations.
1. It cannot police itself (yet).
Because most AI is built on a probabilistic model, there is chance of error inherent in every output that the AI provides. We don’t want to eliminate this error, because that’s what gives GenAI its creativity. We then use repetition and permutation to build on previous responses and refine the answer to reach a non-obvious conclusion.
With each iteration, the level of confidence in the answer will drop, statistically speaking. Even if the AI has a 95% accuracy rate (which is still a pipe dream, by the way), by the time it has gone through three iterations, the confidence level in the response is only around 86% (0.95 x 0.95 x 0.95).
So, when AI hallucinates, who exactly is accountable? It may have come from an error in the source material, a poorly-structured user prompt or simply by chance. In any case, traceability and accountability are an ongoing challenge, and all outputs still need to be verified by human intuition or judgment.
The answer may be to use a separate, deterministic AI to provide a steering wheel that guides GenAI towards a predefined output, but that is still on the horizon and is a topic for another blog.
At this point, the best we can do is look at the output and say “it just smells wrong.” AI can augment and assist us, but humans still have the final say — and the ultimate responsibility to ensure that AI is being used for ethical purposes.
2. AI doesn’t know what you don’t tell it.
Earlier we covered AI’s capability to create personalized training and augmentation for individuals in the digital workplace, but there certainly are limitations. This is a situation where you get as much as you give.
Paradoxically, many of us are what I would call “digital extroverts” in our private lives — sharing personal details via Facebook, LinkedIn and Instagram — while being very introverted about what we share with our employer. And rightfully so.
In a digital workplace every email, chat message or keystroke can theoretically be collected and analyzed. There are obvious privacy concerns and we don’t want our employers spying on us, but this information is a potential goldmine for AI in the workplace.
It can be used to tailor the assistance it provides, the training you receive — even the job opportunities you are considered for. This calls for a culture of trust and transparency in the organization, but how do you find the right balance between privacy and innovation?
Obviously, ethical and privacy guardrails are essential, but it also requires a hyper-personalized approach and a human-centric design. One that provides a higher level of AI augmentation to those willing to share more about themselves, but doesn’t discriminate against digital introverts.
Obviously, AI can bring major benefits to the digital workplace, along with challenges and risks around data privacy, accountability and bias. AI will not eliminate the need for human intuition, communication or empathy, as these are still vital for building relationships, understanding diverse perspectives, promoting the organization’s values and fostering a positive work culture.
For now, AI can only facilitate and enhance our interactions, but we humans still have to take ownership for the decisions we make.
Posted on: May 16, 2024
Janusz Marcinkowski
Digital Workplace Innovation Consultant
Member, Atos Research Community
View details >