Human-AI Collaboration Is the New Teamwork. Ready?

As businesses integrate AI-powered machines into their operations, how humans interact — or should interact — with these enterprise AI systems is revealing. Case in point: When OpenAI CEO Sam Altman, in response to a social media post, remarked that saying “please” and “thank you” to ChatGPT probably accounted for tens of millions of “well-spent” dollars in electricity costs, it set off a spirited debate about whether generative AI users should prioritize politeness or conserve computing power.

The question of how users engage with AI-powered platforms will likely become even more complex as agentic AI — digital agents capable of taking action independently toward a specific goal — becomes more widely available and frequently used.

Companies increasingly recognize that AI’s greatest business value will not come from automating rote tasks, but from augmenting human thinking to solve complex problems. Making this pivot is a business imperative: Today, about eight in 10 companies say they aren’t yet seeing significant bottom-line impact at the enterprise level from generative AI, according to McKinsey.

Much has been written about how the rise of AI will increase the value of uniquely human skills like creative problem-solving or communication. While that’s true, it’s only the beginning of how skills will need to morph and grow to match shifting employer expectations. As businesses become more AI-dependent, humans will not only need to effectively collaborate with one another but also develop new skills to collaborate with AI and eventually manage it

Related:Ways CIOs Can Elevate Task Delegation and Bolster Communication

Whether brainstorming creative solutions, diagnosing patterns in messy data, or assisting with strategic decision-making, AI will increasingly become a collaborator, not just a tool. Sure, getting there will require investing in AI, but more importantly, the new paradigm will require investing in the humans who will use it.

Skills Middleware for AI and Human Collaboration

Middleware in the software context is the connective tissue between different applications or platforms. It allows different platforms to “talk” to each other, breaking down siloed information and creating new ways for existing tools to work together.

Bolstering human skills to support collaboration has long been part of the learning and development domain, most often building on an existing skill set. But when it comes to working with agents as partners, all employees start from square one. The labor force will need to build a skills middleware: the set of competencies that will facilitate AI’s integration into more cognitively complex tasks.

Related:InformationWeek Podcast: Coordinating Crunch Time Across the Company

This skills middleware isn’t just technical proficiency, like prompt engineering or understanding AI model limitations (though those are important). It also includes a more nuanced mix of communication, judgment, ethical reasoning, and task delegation. It requires understanding the strengths and weaknesses of various agentic AI platforms to determine which should be deployed for specific tasks or projects. In essence, it’s the same kind of skill set that good managers use to coordinate high-performing teams — only now, some of those team members might be AI agents.

As AI agents are increasingly able to independently book meetings, search for information, and follow workflows to perform more complex tasks, the business value of AI will increase. It will move from replacing low-cognitive tasks to more complex problem solving, but only if humans have the skills to use AI to augment their decision-making, as well as the skills to effectively manage AI agents.

Developing Skills Through Applied Practice

So, how should companies prepare their workforce for a world where managers oversee not just human teams, but also AI agents?

Employees will need to develop a range of new skills involving truth-finding and discernment. This includes analyzing the outputs of both human-created and AI-developed research and data gathering exercises, such as surveys, experiments or automated data collection by AI systems. Part of this skill set involves spotting hallucinations and algorithmic bias in AI outputs. While AI will be able to spot patterns in large data sets, humans will need to determine which patterns are relevant and meaningful, separating valid correlations from random noise.

Related:AWS’s New Security VP: A Turning Point for AI Cybersecurity Leadership?

These won’t be capabilities that can be mastered through instructor-led training or through a self-serve video library. Developing the middleware skills will need to happen in context, through applied practice.

The Case for Building AI Skills Through Volunteering

As I know through my work at the Taproot Foundation, a powerful, yet often overlooked avenue for gaining applied practice is through skills-based volunteering. By deploying their professional expertise to help nonprofits and social enterprises, employees can gain invaluable expertise.

These projects often operate in a constrained environment, limited by both budget and time. The constraints may make AI an important part of a project, giving volunteers a chance to develop new workflows or other solutions that take advantage of AI to help meet the nonprofit’s mission. 

Another benefit? Pro bono work offers applied learning with real-world consequences and higher stakes, which is closer to the experience of a job than project-based learning rooted in a hypothetical scenario or a low-risk, low-reward bonus project that isn’t core to the success of an organization.