No other technology in recent history has received as much attention as artificial intelligence. It's in the press, social media commentaries, at conferences, and in general conversations. It's everywhere — but how much of it is real versus hype?

"It's such a tiny word, but it covers quite a lot of different disciplines," says Adrian Wood, director of strategy and marketing for Dassault Systèmes. "However, I think the term AI is overused and misunderstood quite a lot. In general, AI can be defined as machine algorithms acting in the way that humans do. It's trying to mimic intelligence artificially. There are a lot of different aspects beneath that simple understanding,”

Indeed, businesses are adopting AI technology at a rapid pace. According to an annual survey by McKinsey & Company, 78% of respondents indicated that they use AI in at least one business function. In comparison, 20% indicated the same in 2017. 

Evolving AI

According to Wood, AI is considered an umbrella term and has evolved over the years. Traditional AI dates back to at least the 1960s and encompasses techniques such as machine learning, which utilizes historical data to optimize problems and predict future outcomes.

There are also what Wood describes as "the new kids on the block" — generative AI and agentic AI.  Apple's Siri and Amazon's Alexa are personal-use examples of these, but from a business context, their uses are just starting to be implemented. 

MIT defines generative AI as a machine-learning model that is trained to create new data rather than predicting a specific dataset, whereas agentic AI refers to AI systems that can autonomously take actions, make decisions and adapt to new situations without constant human intervention.

Generative and agentic AI are what Wood considers the peak of the AI hype cycle. "There are a lot of companies that are investing in this technology,” he says. “But I'm not sure whether companies understand exactly what it does and how they can apply it for real business value."

Despite this lack of understanding, a 2024 McKinsey survey found that 71% of companies were using generative AI in at least one business function. That's up from 65% just six months earlier. 

AI Challenges

As the adoption of AI continues to grow, companies must be cautious when utilizing next-generation AI. These AI solutions can generate information that's incorrect, misinterpreted or otherwise flawed. According to Wood, that's because, in concept, generative AI is basically just a probability machine. That is, it's looking for the most likely next word that should follow in a sentence. 

"If we put that into a business context and start asking generative AI or agentic AI to make a major decision about how to reorient our supply chain, that is incredibly risky," says Wood. 

Additionally, AI requires both structured and unstructured data to drive operations. Data that is clean, uniform and ready to process is a requirement. However, the challenge arises when the data has a hidden bias that humans don’t recognize. If there’s bias, AI will produce a biased output. As a result, users may receive decisions that are not objective in terms of identifying specific suppliers, customers, manufacturing facilities or distribution methods.

Solving AI Challenges

To solve such challenges, companies should implement a strong data-governance framework. Resources include the U.S. Department of Commerce's National Institute of Standards and Technology (NIST) which developed a voluntary-use framework designed to incorporate “trustworthiness” into the development of AI products, and the Joint Development Foundation Project’s Coalition for Content Provenance and Authenticity (C2PA) which has developed technical standards to certify the provenance of media content. 

Additionally, AI algorithms must be designed to minimize bias and ensure fairness in decision-making. Companies should also regularly audit AI models to detect and mitigate biases, ensuring that AI-driven decisions align with ethical business practices.

All of this involves skillsets that businesses will need to add. "Companies will need to identify and add new roles, responsibilities and governance to make sure they comply,” Wood says. “This can be challenging for many smaller and mid-sized companies."

With that being said, Wood emphasizes that companies are already deriving significant value from traditional AI today, such as optimization and machine learning.  "It's the backbone of many planning organizations." 

Like second-generation AI technology, traditional AI also requires discipline and data requirements to drive it. But the risk is not as high.

Building Trust

One of the major concerns is understanding the decision-making process that leads AI to generate a specific answer or recommendation. A lack of transparency, known as a “black box” situation, occurs when machine learning models learn from vast amounts of data, resulting in complex internal structures that make it difficult to trace their decisions.

As such, companies need to ensure and build trust in their AI models.  First, employees must be educated in basic AI operations, ethical use of the technology and responsible data management. When individuals understand the nuances of AI, they interact with it more effectively and confidently.

To test AI within an organization, Wood suggests starting small, by picking a pilot project where there’s a known level of high data maturity. Next, set a specific goal with measurable milestones, and be sure to get executive buy-in and support.

For companies that are in the early stages of digital transformation, and still dependent on spreadsheets and legacy systems, traditional AI works extremely well, according to Wood. It's proven and it provides value. 

Wood says companies must first be moved away from manual and reactive processes to more automation and supply chain optimization. Companies that have been using AI for a while and are asking, “What's next?” can test generative AI via small projects. The goal is to provide companies with scientific precision based on virtual twins of their supply chains and planning, enabling them to make informed decisions.

Outlook

The AI train has left the station, and there's no turning back. Companies are eager to capitalize on the promise of AI, but this enthusiasm must be tempered with a clear-eyed view of the risks. For those still in the early phases of digital transformation and relying on spreadsheets or outdated legacy systems, traditional AI solutions offer a stable and proven path forward. 

As the industry advances, the next few years are expected to see an increase in the adoption of generative and agentic AI. At the same time, they will need to address continued challenges: hallucinated outputs, biased decision-making, data-privacy issues and the lack of transparency in how conclusions are reached. 

Companies will need to implement powerful governance frameworks, invest in training, and ensure transparency in AI-driven processes. As AI systems become more capable and autonomous, trust in the technology — and in the people deploying it — will be the deciding factor between scalable success and costly setbacks.

Resource Link: https://discover.3ds.com/right-supply-chain-planning-intelligence