The AI Teammate Revolution: What P&G's Groundbreaking Study Means for Documentation Teams
Worried about AI taking your job? Discover how new research shows AI is a 'cybernetic teammate' for knowledge base teams, boosting performance, sharing expertise, and making work more fun.
If you've been on LinkedIn lately you've probably seen AI everywhere. It's usually hyped up, showing how AI is better, super useful, and can be twisted into just about any role.
We've all heard the buzz about Generative AI (GenAI), and yeah, it's pretty much everywhere now—from automating boring tasks to sparking cool new ideas.
But let's be honest, for many of us who work with knowledge bases and customer support, AI talk often comes with a side of worry.
Will it take our jobs? Is it just another tech thing to learn? Does it even get our product? These are totally fair questions, and seriously, we hear you! Here's the kicker: when you actually try to use AI, it can sometimes be a bit... well, underwhelming.
It might lack context, suck the creativity right out of your content, and honestly, all the AI chatter out there can just get plain overwhelming 😵
Now, I know this might sound a little weird coming from us, especially since we have AI built right into our product. The truth is, I'm not saying it's not useful—I'm just saying it doesn't have to do absolutely everything for us.
And that's exactly why we've always told technical writers to use it for the dull, tedious parts of their jobs. Think generating meta descriptions or getting those first article drafts going, making it way easier to kick off a new help article.
But guess what? A new research paper from Harvard Business School, called "The Cybernetic Teammate," just dropped some seriously insightful findings that every single documentation writer needs to hear.
The researchers ran a huge experiment with 776 pros at Procter & Gamble, and what they found is both incredibly promising and, honestly, a little concerning.
Today, we're taking a super deep dive into what this research means for the future of knowledge bases, how we can totally rethink our AI goals, and even tackle some of those nagging worries head-on.
Let's break down the whole story: the good bits, the tricky parts, and the "we totally need to figure this out" aspects.
Beyond the Tool: AI as a True Collaborator
For ages, we've thought of AI like a fancy calculator or a super-fast spreadsheet—just a tool to make us perform better. But Large Language Models (LLMs), the real stars of GenAI, act way more like a person than a machine. Why? Because they're trained on human language!
This got researchers from Harvard Business School, Wharton, and Procter & Gamble thinking: Can GenAI actually fill a human role in teamwork? Can it bring some of those classic benefits, like boosting collective performance, sharing expertise, or even making social connections?
As the authors put it:
"Prior work has treated Al primarily as a tool, like a spreadsheet or calculator, that can be used to enhance performance. But a unique aspect of Large Language Models, the most common form of GenAI, is that they are trained on human language and often act more like a person than a machine (Mollick, 2024). This leads to a key question: can GenAI fill the role of humans in teamwork?"
They set up a massive field experiment with 776 professionals at P&G, tackling real product innovation challenges. Participants were split into four groups: individuals without AI, teams without AI, individuals with AI, and teams with AI.
The results? Seriously eye-opening!
The Performance Power-Up: Doing More, Better, Faster (But Are We Losing Something?)
One of the most interesting findings is just how much AI can boost performance.
Get this: individuals who worked with AI produced solutions that were just as good as what two-person human teams without AI could manage. That's a massive deal!
The research plainly states: "Individuals with Al produce solutions at a quality level comparable to two-person teams, indicating that Al can indeed stand in for certain collaborative functions."
It suggests AI can really step up for some collaborative tasks, acting like a genuine teammate. It gives individuals access to all sorts of expertise and viewpoints that you'd usually only get from a team.
So for your knowledge base team, this isn't about one person doing the work of two. It's about one person, powered by AI, having the capacity to hit the quality levels that used to take a whole team working together.
What does this mean for writing documentation? Let's take a peek.
Faster Content Creation
The study found that individuals using AI spent a whopping 16.4% less time on tasks, and teams with AI spent 12.7% less time. On top of that, AI led to way longer outputs.
As the paper notes: "the introduction of AI substantially reduced time spent working on the solution: individuals with AI spent 16.4% less time than the control group, while teams with AI spent 12.7% less time."
For knowledge base writers, that means drafting more comprehensive articles in less time. Imagine starting a new topic with an AI-generated draft that's 500 words longer than you'd usually crank out. That's a ton of time saved!
These perks were even more noticeable for non-native English speakers, suggesting a "leveling effect" that helps close those performance gaps.
Reduced "Starting from Scratch" Syndrome
Remember how we talked about AI making the boring stuff just disappear? This is totally it!
AI can handle the initial heavy lifting of content generation. That means writers can skip staring at a blank page and jump straight into polishing, optimizing, and adding that special human touch.
Many technical writing tools already show how AI can make content creation smoother, generate first drafts, and automate tedious tasks like formatting and citing sources.
Now, a quick but important heads-up: the folks in this study were pretty new to AI prompting. This means the benefits they saw might actually be less than what's truly possible as users get smarter with their AI interactions.
So, the more we learn to chat with our AI teammates, the more awesome they'll become!
Breaking Down Silos & Democratizing Expertise (While Potentially Creating New Dependencies)
One of the most exciting things for knowledge bases is how AI can smash down functional silos and make expertise available to everyone.
The research pointed out a common problem in organizations: without AI, R&D pros usually suggested technical solutions, while commercial pros leaned towards business-focused ideas.
Sound familiar? In knowledge bases, this often means content that's either too techy for the average user or too vague to really help.
But here's the magic: pros using AI churned out balanced solutions, no matter their background.
The study says: "Without AI, we observed clear professional silos - Commercial specialists proposed predominantly commercial solutions while R&D professionals favored technical approaches. When teams worked without AI, they produced more balanced solutions through cross-functional collaboration. Remarkably, individuals using Al achieved similar levels of solution balance on their own, effectively replicating the knowledge integration typically achieved through team collaboration."
AI helped them "reason across traditional domain boundaries and approach problems more holistically." Basically, it helped them think outside their usual boxes.
What does this mean for our knowledge base teams?
Richer, More Balanced Content
Imagine a knowledge base article about a tech feature. Instead of just the engineering view, AI can help mix in user experience ideas, marketing angles, and common customer questions. This makes the content super comprehensive and easy for way more people to understand.
AI-powered knowledge management systems can also spot patterns in messy data (like ticket comments) to whip up comprehensive knowledge articles, making your content repository even richer.
Empowering Everyone to Contribute
The study found that "non-core-job employees" (those who weren't as familiar with new product development) hit performance levels similar to teams with at least one core-job employee when they had AI.
This is massive for knowledge bases! As the researchers explain: "AI allows less experienced employees to achieve performance levels that previously required either direct collaboration or supervision by colleagues with more task-related experience."
This means AI can effectively stand in for the deep expertise and guidance you usually get from experienced team members.
So less experienced writers, or even subject matter experts from other departments, could contribute high-quality content way more independently, without needing hand-holding.
Finding Content Gaps & Ensuring Context
That whole "it needs too much context" or "it doesn't get our product" problem often comes from a lack of integrated knowledge. AI, acting as a bridge, can help you find where your knowledge is thin or inconsistent across different areas.
It can even prompt you for missing info or suggest how to rephrase tech jargon for a more general audience. This makes it easier to fill those crucial content gaps and provide the context everyone needs.
Technical writers are even becoming "information architects" now! They're designing AI-ready knowledge ecosystems that deliver smart information experiences, moving beyond just writing user-friendly docs.
The Emotional Connection: AI as a Positive Presence
Here's a surprising one, especially if you've ever felt frustrated trying to get a machine to understand you: participants who used AI reported more positive emotions (like excitement, energy, and enthusiasm) and fewer negative emotions (like anxiety and frustration) compared to working alone.
This is a big deal because past research often said tech had a negative impact on social vibes at work. But this paper totally backs it up, stating: "Contrary to fears about Al creating negative workplace experiences, we found consistently positive emotional responses to Al use, including increased excitement and enthusiasm, as well as reduced anxiety and frustration."
This "positive emotional experience" with AI fits right in with new evidence that conversational AI can actually boost people's social and emotional well-being by giving encouraging, human-like responses.
In fact, other studies hint that AI-based writing tools can make you feel more confident by easing anxiety and clarifying requirements. They can even help human customer service agents sound more empathetic and thorough.
"AI is boosting what individuals can do, freeing up humans to focus on tasks that need critical thinking, empathy, creativity, and big-picture oversight—stuff AI still can't really do."
For knowledge base teams, this is fantastic news:
- Reduced Stress and Burnout: Imagine spending less time battling writer's block or feeling swamped by a complex topic. If AI can cut down on that friction and make the experience more positive, it can lead to happier, more engaged writers.
- Increased Job Satisfaction: If AI takes care of the "boring stuff" like generating first drafts or metadata, writers can pour more energy into the creative, strategic, and problem-solving parts of their roles.
What About the Negatives of AI in Documentation?
While the research generally paints a rosy picture, it's super important to talk about the challenges and trickier parts of bringing AI into the mix. It's not a magic bullet, and understanding the potential pitfalls helps us navigate the future a lot more effectively.
The "AI will take my job" Fear
This is a very real and totally understandable worry. The study's finding that AI-powered individuals can match the performance of human teams might, at first glance, fan the flames of this fear.
But here's the key takeaway: it's about augmentation, not replacement.
As some researchers put it, "AI will, however, transform knowledge work, and people will still be vital for making decisions, being creative, and solving problems."
AI is boosting what individuals can do, freeing up humans to focus on tasks that need critical thinking, empathy, creativity, and big-picture oversight—stuff AI still can't really do.
Our job shifts from just churning out content to being content strategists, curators, and quality assurance pros. Pretty neat, right?
The Hallucination Headache & Trust Deficit
This is a huge one for knowledge bases, where being accurate is absolutely critical. AI, especially GenAI, can sometimes "hallucinate"—meaning it makes up plausible-sounding but totally wrong information.
The research highlights that while objective performance got better with AI, participants actually felt less confident about their solutions.
As the paper points out: "Interestingly, while objective performance improved, participants using Al were actually less confident about their solutions. [...] Al-enabled participants were 9.2 percentage points less likely to expect their solutions to rank in the top 10% compared to the control group (p<0.05), suggesting a disconnect between actual and perceived performance."
River Sloane
This "disconnect" means that even though AI can be powerful, human oversight and double-checking are absolutely critical. We can't just blindly trust what AI spits out. Building a workflow that includes rigorous fact-checking is a must!
"Doesn't Understand Our Product" & Contextual Limitations:
While AI can help bridge knowledge gaps, it doesn't really "understand" things in the human sense. It just relies on the data it was trained on and the prompts you give it.
If your product is super niche, really complex, or constantly changing, AI might struggle to give you truly accurate or nuanced context without a lot of human input and fine-tuning.
AI is "prone to errors when it encounters data that's different from what it was trained on," and it might "fail to comprehend the situational context, and even hallucinate nonexistent product information." This just reinforces that subject matter experts must guide and correct AI, especially for highly specialized knowledge.
Potential for New Silos and Homogenization
While AI helps break down old silos by balancing output, if it's not integrated well, it could accidentally create new ones. If teams become too dependent on AI without keeping up human collaboration, there's a risk of less human interaction and a weaker ability to organically share knowledge.
The "team + AI" scenario still had the best shot at producing top 10% solutions. This suggests that the combo of human collaboration with AI is where the truly amazing stuff happens.
It's not AI instead of teamwork, but AI with teamwork.
River Sloane
The study notes: "While individuals with Al show a small positive effect, this effect is not statistically significant, suggesting that the combination of AI and teamwork might be particularly powerful for achieving exceptional performance."
Plus, AI-aided solutions in the original study showed "notably higher semantic similarity."
That means they were becoming more alike. This is a bit worrying because it could mean we lose the diverse perspectives and unique voices that make documentation truly outstanding.
Over-Reliance and Skill Atrophy
The study found that many participants kept 75% or more of the AI-generated content in their final solutions. While this shows AI is effective, it also makes you wonder about just passively accepting what AI gives you.
There's a risk of "workforce skill degradation, with team members relying too heavily on AI for decision-making, reducing critical thinking and problem-solving skills."
Are we building new skills or just building a new dependency? Something to think about 🤔
Implementation Challenges
Bringing AI into knowledge management comes with some real-world bumps. Setting up AI-powered KM systems can be "time-consuming and costly," needing a big initial investment.
Integrating AI into old, outdated IT systems can also be a huge headache. And making sure your data is high quality for AI training, plus handling data privacy and security, is super important.
The Cybernetic Knowledge Base Awaits 💫
The "Cybernetic Teammate" research truly changes how we see AI at work. It challenges the idea that AI is just a fancy search engine or a handy text generator. Instead, it shows AI as an active player in team collaboration.
"By embracing AI as a collaborator—a teammate that can handle the boring stuff, boost our performance, break down knowledge silos, and even make our work more fun—we can build more reliable, up-to-date, and ultimately, more human knowledge bases."
For knowledge base teams, this means a big shift: instead of just slaving away writing every single word, we get to be strategic guides for AI, curating its outputs, and making sure our knowledge bases are not just comprehensive, but also super empathetic and genuinely helpful.
The authors perfectly sum this up: "Al’s role transcends that of a mere tool or facilitator, entering the relational fabric of collaboration itself. By treating Al as an active counterpart, and in fact as a proper teammate, we gain deeper insight into how GenAI mediates, and is mediated by, the collective processes that form the backbone of modern teamwork."
By embracing AI as a collaborator—a teammate that can handle the boring stuff, boost our performance, break down knowledge silos, and even make our work more fun—we can build more reliable, up-to-date, and ultimately, more human knowledge bases.
What an exciting time to be in documentation! 🌟