How AI Can Transform Business With Decision Making 2.0

Soumendra Mohanty is the Chief Strategy Officer at Tredence Inc.

Over the past few decades, the approach to decision making has evolved as a result of many technology innovations and business model changes, such as systems of record, engagement and intelligence, as well as software as a service (SaaS). Metaphorically, this is the paradigm of “decision making 1.0,” where humans input data into systems and tools, interpret insights and recommendations and then make decisions based on that output.

As much as we may believe that humans weigh all the options and always make rational decisions, the reality is far from that. One reason is that the human brain’s cognitive capacity is limited, so it simply can’t process the vast information needed for certain decisions. Or it has certain biases, so it just decides which option is “good enough.” But that may not necessarily be the optimal decision. This is called “satisficing.” This happens when you’re overwhelmed with choosing a restaurant from hundreds of options or when a CEO is making a complex business decision.

Another reason for human limitations is that businesses deal with many challenges, competing priorities and various alternatives, with each one presenting a paradox. So, it’s difficult for humans to parse all these competing options.

Decision Making 2.0

Generative AI can quickly interact with data, test hypotheses, simulate scenarios and provide summarized insights. AI agents can now be embedded in the workflow, accelerating actions autonomously. This is a paradigm change, enabling much better decision making beyond just satisfactory decisions. We can term it as “decision making 2.0.” Tools and agentic systems not only comprehend the environment as it changes, reason and break down the problem into a chain of tasks, but they also execute the tasks. Humans play an oversight and observation role to provide feedback for continuous improvement.

This new era of decision making affects many levels of an organization. At the managerial level, AI assists in the day-to-day, tactical routine decisions in a faster and more informed way. This could extend across many verticals. For example:

• In retail, AI can assist in managing stock and optimizing campaigns based on past performance and market trends.

• In CPG, AI can help brand and category managers make data-driven decisions.

• In pharmaceuticals, AI can help sales representatives visit with doctors based on potential prescription usage.

• In banking, financial services and insurance, AI can give real-time wealth management and investment guidance.

At a strategic level, CIOs and CXOs can also benefit from generative and agentic AI systems that can assist with narratives concerning internal business performance, as well as external aspects such as market intelligence, competitor strategies and future outlooks. This allows them to make more holistic, comprehensive and optimal decisions.

The Emergence Of Co-Intelligence

The growth of generative and agentic AI is just beginning. It’s now focused more on general operational and tactical, repeatable tasks that can be autonomously completed. But soon, deeper vertical specializations will emerge to address more real-world problems. In this new world, we’ll have “co-intelligence.” This will involve humans who are experts in certain areas, and they’ll have agents surrounding them. The agents will become more autonomous, but humans will still play a critical role in governance, providing reinforcement learning and feedback, observing the performance and defining boundaries, guardrails and policies.

There are key aspects of generative AI and agentic AI that make these tools easier to adopt for decision making. Previously, using AI required a certain level of technical prowess to engage with algorithms. People had to bridge the gap between technical outputs and business insights. But now they’re much easier to interact with due to their natural language conversational capabilities, which breaks the barriers between business language and technology language.

Building Trust And Adoption

Still, for companies to truly adopt and benefit from generative and agentic AI systems, a gradual, structured approach is required. A barrier to adoption is the fear of losing control, especially for people with deep knowledge of their domains. Companies should introduce AI as an assistant, with a human in the loop, with reinforcement learning to provide guidance.

Humans may take time to get used to the speed of AI, but the benefits of this speed can be significant. Whereas humans previously could require several meetings to pore over numerous datasets and reports, trying to grasp a situation and devise decision steps, AI tools can simulate scenarios in seconds and make real-time recommendations to aid in decision making.

Another aspect of trust and adoption is the concern with how AI makes decisions—the so-called “black box effect.” To address this, companies should develop AI fluency to understand what AI does and how it works. Large language models (LLMs) make predictions in the form of a probabilistic output after observing patterns in vast amounts of data. Therefore, it’s important to have responsible AI frameworks, guardrails and policies so that people can understand how the AI is performing, where it’s failing, what the implications are and what can be done to improve. For example, guardrails could help ensure data sources are valid to protect against misleading or biased results. There are also many frameworks and transparent AI benchmarks for better evaluating and understanding models.

Developing and implementing a foundational AI governance and ethics framework for an organization is critical. Oversight at all stages, including ensuring appropriate data, as well as building, training and testing models, is important. A responsible AI framework includes continuous monitoring and testing for accuracy, hallucinations and other metrics. Developing a cross-functional team consisting of data, tech, legal and HR leaders who examine outputs is recommended. This team certifies that a model is good to be released. Other strategies, such as a champion-challenger approach in which different models compete in providing the best answers in terms of fairness, reliability and compliance, can be adopted.

Conclusion

CIOs in particular should pay attention to this shift toward decision making 2.0, as the technology industry will see these changes faster than other sectors. By leaning into this new form of decision making, you can reshape business workflows and competitive strategies. Ultimately, embedding AI into workflows will help you make better decisions faster and avoid pitfalls.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?