Gen AI’s ultimate potential? Hive mind teamwork

Under a sun-bleached tent outside Baghdad, we came to a grim conclusion: Our elite counterterrorism teams (JSOC) — composed of the best-equipped forces in history — were losing to an enemy armed with tech you could fish out of a bargain bin.

This enemy could form a plan, execute it on the battlefield, and be onto their next target before we knew what was happening. 

Yes, our communication tech was far more advanced, but our problem was friction. By the time information slogged through every layer of our hierarchical operating model, all comparative advantages were lost.

The enemy’s model, however, amplified the power of their bargain-bin tech. They could push information to anyone, anywhere, and at any time. They could execute quickly, adapt on a dime, and strike with devastating effect.

It was like fighting a dynamic organism — a constantly evolving network of cells that seemed to think and act as one.

Under that tent, we weighed our options: Fundamentally change the way we’d operated for decades or lose.

I believe large language models (LLMs) will prompt most organizations to face the same reckoning.

LLMs are already revolutionizing business functions but their true potential lies in accelerating strategy execution to speeds unimaginable today. The catch? Organizations with execution processes built for speed will be positioned to gain the most from LLMs. The rest risk falling too far behind to catch up.

The true promise of LLMs

Leveraging deep learning and enormous data sets to analyze, understand, predict, summarize, and generate natural human language, LLMs bridge the gap between the latest AI and the bargain bin. Virtually anyone can tap into their ability to instantly analyze information and respond intelligently in plain language. 

Bets are already being made on LLMs to revolutionize customer experience, marketing, research and development, and more. But if you think about the time and energy we spend coordinating, correcting, inquiring, sifting, analyzing, sending, and responding to information vital to a shared goal, you can begin to visualize LLMs’ greater potential.

For JSOC, accelerating that process was key to accelerating our execution. Our solution? A meeting we called the Operations and Intelligence (O&I) brief. Every day at a set time, we met to “load up” our brains with the most current insights, put them to work for the next 24 hours, and then reconvene to do it all over again.

The O&I is standard military practice, but what made this brief different was that every JSOC member or partner was invited, no matter their role or location. What’s more, everyone was expected to share all their intelligence. 

This went against decades of cultural conditioning, so at first, everyone hated it. Still, it became the heart of the organism we sought to create and the pulse of our operations. And it worked. The more our teams communicated with each other across the hierarchy, the more everyone knew what everyone else was doing, how it all fit together, and how their information could help.

We called it “shared consciousness,” and the more it grew, the better our decisions got and the faster we could act on them. Over time, executions that previously took days went down to hours. For us, it was a record-breaking level of speed.

Organizations that leverage LLMs will make it look glacial by comparison.

As vital as our O&I was, it had limits. It took us away from the field and execution. Time constraints made it difficult for everyone to contribute. Information aged quickly and often changed after we were dismissed. Intelligence received before or after the O&I often couldn’t be integrated into our full operations until we convened again.

LLMs will enable a shared consciousness without those limits:

Knowledge sharing. Unlike a static database, LLMs function more like a brain. They get smarter as more information is fed to them. They can also use that information to improve themselves. Not only can anyone access exactly the information they need anytime and from anywhere, but the LLM will also be able to provide plain-language clarification and suggest additional resources in nanoseconds.

Scalability. LLMs can accommodate growing teams without limiting anyone’s ability to participate, seek assistance, or provide input.

Updates. LLMs can mine the queries and clarifications they receive from users to absorb and distribute the most current information.

Toward the ‘hive mind’

How might this notion of LLM-enabled shared organizational consciousness look in practice?

Imagine it’s 2028. You’re reviewing a message before it’s broadcast to the organization. A bot appears on your screen and says, “Carol just suggested the VPs each get this message directly from you. That seems wise, given the timeline. Do you want me to write those up and personalize them?” “Yes,” you say.

The messages appear, and your bot announces that Nick is requesting a video call. “Would you like me to take that call for you?” the bot asks. “Please,” you sigh in relief.

While you review the messages, your bot takes the call and returns with another suggestion. “Nick has fresh numbers on the Henderson deal. Would you like me to incorporate them into the VP messages and alert Carol?” “Yes.”

There was a time when this scenario would have seemed far-fetched. Not anymore.

Like JSOC did years ago, I believe most organizations will soon find themselves up against “hive minds” — competitors using LLMs to think and act as one at incredible speeds.

Fortunately, we were able to catch up to our adversaries, but I think organizations today have only two options: Get in front of this power curve now or become irrelevant. LLMs are improving too fast for anything in between.

Take action

Here are three actions for organizational leaders to consider as they navigate the potential of LLMs.

1. Test the limits of LLMs: It’s hard to imagine the benefits (or threats) of LLMs if you don’t know what they can do. Find ways to incorporate them into your work and that of your teams to get as many perspectives as you can. Build prototypes to explore opportunities around messaging, calendar management, coordination of tasks, and more. (I know a CFO and team who build one chatbot every week.)

2. Embed transparency everywhere: The technology won’t be enough. We learned that the hard way. In fact, the key was transparency, which we would not have been able to maintain without relentless discipline. As powerful as they are, LLMs will still be at the mercy of your teams’ willingness to share their information. If that sounds like a tall order for your organization, a cultural change is likely needed, and it must be led from the top.

3. Develop LLM guidelines: LLMs and the AI that powers them are not perfect. Concerns such as data bias, privacy, the accuracy of output, and what to do about them continue to be debated. As governments zero in on how to regulate this technology, set boundaries on how your organization’s data can be utilized. Know how to decommission projects and delete data if necessary. Establish protocols for ethical use and monitoring, and be prepared to update them as regulations evolve.

As LLMs continue to improve, organizational leaders face a critical decision: Embrace this transformative technology and stay ahead of the curve, or risk becoming irrelevant. There is no reason to think there will be a third option.

Chris Fussell’s colleague Will Smith also contributed to this article.