Robb Wilson, Co-Founder and CEO of OneReach.ai and co-author of the WSJ bestseller Age of Invisible Machines says the advance of Artificial General Intelligence (AGI) is an opportunity to make ‘intelligent digital workers’ trusted and revered teammates.
Artificial General Intelligence is a touchy term.
What AGI really points to is an AI system that mimics human intelligence, possessing cognitive flexibility and problem-solving capacity.
AGI can understand things, learn new things and work across a wide range of tasks and domains.
A nascent form of AGI is no longer the realm of science fiction, though most investors and organizations remain unaware.
While there’s been plenty of frenzied activity around acquiring generative AI tools that can bolt on to existing systems, a handful of forward-thinking organizations are on a different path.
By creating technology ecosystems where LLMs are used to orchestrate existing software conversationally, they are on the journey to organizational AGI.
The tools and strategy necessary for creating Organizational AGI are currently available.
Enterprises are already using conversational AI as a thin UI layer, allowing employees and customers to interact directly with digital teammates that I call intelligent digital workers (IDWs).
These IDWs can use all forms of data across an organization—often leveraging generative AI to mine unstructured data.
IDWs can draw from a shared library of skills that can be customized to create personalized experiences.
Whether building from scratch or restructuring existing systems, creating the requisite ecosystem for this kind of activity requires a lot of work.
I’ve spoken with far too many CIOs who think that giving their workforce sanctioned access to an LLM will open the floodgates of automation.
We’ve already seen that people are indeed leveraging the strengths of generative AI to write emails, summarize content and test code, but these are relatively limited use cases in terms of the kind of productivity enhancement that conversational AI can deliver with the proper long-term strategy in place.
Multiple enterprises have implemented such strategies and seen firsthand the immediate effect of getting everyone within an organization involved in the process of designing and evolving automations.
This strategy also has the long-term effect of fostering a budding awareness of an organization within the technology ecosystem, setting the stage for conversational AI to survey all the data and read or listen to all forms of communication.
The technology ecosystem that’s grown within the organization maps this information to skills that lead to improved experiences with employees and customers.
From minute to minute, IDWs can know what the organization knows, eventually understanding its operations along vectors that humans might not even be aware of.
This is the organizational AGI I see on the horizon.
None of this is to imply that organizational AI will suddenly realize its survival depends on eliminating humans.
In fact, the formative activity within an ecosystem that’s growing toward organizational AGI is human-led and human-fed.
The skills that IDWs run are created by humans.
Ideally, they are created by humans who best understand the tasks they are seeking to automate.
These humans create these skills using low- and no-code design tools that essentially enable them to write software conversationally.
Humans can come up with ideas for automations, quickly build frameworks for the skills that will execute the automations, test these skills and iterate on them.
Most importantly, humans and IDWs are in continual contact via human-in-the-loop (HitL).
This powerful design pattern has IDWs turning to humans when they don’t know how to proceed or in moments where one human needs to interact with another.
The ongoing process of HitL provides critical guardrails for the broad adoption of AI within an organization.
HitL is also fundamental to training these systems.
With humans designing and maintaining the skills that these AI systems execute, most of the reasoning and decision making will be deterministic – decisions will be predetermined by humans making decision trees.
The agency available to the machine will be very narrow – so it’s a stretch to think that AI systems established by companies will bring us to an extinction-level endgame.
The ability for AI systems to perform actual reasoning and decision-making has come closer over the past few years, but it is still far from being a reality.
As more organizations restructure themselves around conversational AI they will get more work done and the AI’s ability to automate increasingly sophisticated tasks will make IDWs trusted and revered teammates.
Still, even as these systems mature, decision making should remain human led.
Both strategy and reason should be top-of-mind for massive corporations and headstrong upstarts alike.
Conversational AI could lead us to a world where abundance is shared, productivity is enjoyable and people live as equals in relative harmony.
Putting profits and shareholders ahead of the well-being of employees and our home planet is a dangerous trajectory to maintain as these technologies take root in our lives.
Many of the structures and goals assigned to ‘business’ at large simply aren’t sustainable and applying AI to them seems likely to hasten our demise.
On the other side of this, organizations that use conversational AI to improve experiences for their employees first can unleash a new paradigm for productivity that will fuel the automations they share with customers.
If AGI is going to emerge from a decentralized, widespread system of computing resources it seems possible that significant portions of it will emerge from within companies that have achieved organizational AI as they communicate with and learn from one another.
This means that the organizations that get it right in the coming years will have a massive impact on the AGI that emerges in a decade or so.
In one respect, getting right means adopting a holistic strategy that brings AI into all aspects of your organization, not collecting piles of bolt-on tools.
Getting right also means thinking carefully about what it is you want to do with these powerful machines that are learning evermore by the minute.
