Are AI Agents taking over our jobs?

When I woke up this morning, I opened my email and saw an interesting article by Joe McKendrick, a?contributing writer for?ZDNET . In his article "AI agents are the next frontier and will change our working lives forever," he refers to a recent McKinsey Quarterly article, "Why agents are the next frontier of generative AI ."

We have all been bombarded with news of LLMs, AI, and Generative AI and how they will change our lives. We have also seen articles saying that there are lots of organizations that have poured money into AI and are now pondering why they can't see the return on investment. Organizations and humans expect to get an immediate benefit from any investment we make, but that is unfortunately not necessarily the case with AI.

Investing in AI can mean many things to many organizations. To some, it is a fundamental shift in the business model. To some, it is a shift of smaller automation, and we have yet to discuss AI. I have written several articles on the pendulum of complexity that AI represents, and the acceleration of development is just increasing. Every company needs to figure out where they want to be on the "complexity pendulum," but everybody has to start to avoid being left out of the development.

Having solutions such as Microsoft 365 Copilot is just a start for office workers, the way we do workflows today will be completely different tomorrow. The question is when one should "jump in" to ensure that the organization and the employees are trained in the new ways of work.

The ZDNet article offers interesting views and highlights how Generative AI can bring agent-based systems to the forefront. McKinsey defines agentic systems as "digital systems that can independently interact in a dynamic world." That sounds both too good to be true and a frightening concept.

If you have built any chatbot or workflow logic (I have), you know that it might be a horrifying experience as you have to build all of the potential logic into it. However, by including a foundation model in agentic systems and by having these models trained on large and varied unstructured data sets (and not on predefined rules), these agents can adapt to different scenarios in a similar manner as LLMs can respond to prompts.

According to Lareina Yee from the McKinsey team, "We are at the beginning of an evolution from knowledge-based, gen-AI-powered tools, where chatbots can answer questions and generate content to Gen AI-enabled agents that use foundation models to execute complex, multistep workflows across a digital world."

If you consider it, hundreds of millions of "hard-coded" automation rules built over the past years might eventually be replaced by self-directed AI agents that make decisions about the best outcome in the flow. I am sure that in some instances, we don't want to let the agent make a decision, such as in medical emergencies where we still want to have a pair of eyes looking at what the agent did.

The ZDnet article also refers to a recent Capgemini survey of 1,100 tech executives. According to the study, 82% of these 1,100 executives intend to integrate AI-based agents across their organizations within the next three years. Furthermore, 70% of the respondents said that they would trust an AI agent to analyze and synthesize data, and 50% would trust an AI agent to send a professional email on their behalf. 75% said they intend to deploy AI agents to tackle tasks such as generating and iteratively improving code. Other potential tasks for agents included generating and editing draft reports (70%) and website content (68%), email generation, coding, and data analysis.

The Zdnet article refers to a set of different types of agents (check out from the article):

??Simple reflex agents

??Model-based reflex agents

??Goal-based/rule-based agents

??Utility-based agents

??Learning agents

??Hierarchical agents

Yet again, the list shows that an organization can go from simple agents, such as resetting passwords, to hierarchical agents, where agents are in charge of other agents.

There are lots of organizations that still need to start their journey to automation or discussion of AI, and now we are already talking about agents talking to other agents. Decisions are made based on the best outcome defined by the agent by using a complex reasoning algorithm. All this development will unfortunately also increase the confusion on the market and potentially even fear for some organization to take the stance of "wait and see". I have news for these 'wait-and-see" organizations: it might be too late for you to get started if you wait too long, as your competitors are probably already working on changing their way of work or even business models.

I would love to hear your thoughts on this topic.

Yours,

Dr. Petri I. Salonen

PS. If you would like to get my business model in the AI Era newsletters to your inbox on a weekly or bi-weekly basis, you can subscribe to them here on LinkedIn https://www.dhirubhai.net/newsletters/business-models-in-the-ai-era-7165724425013673985/