Back to home

What businesses are getting wrong about agentic AI

The next phase of the AI revolution is coming. WPP’s Chief AI Officer Daniel Hulme explores how businesses can cut through the hype and strategically prepare for the transformative potential of agentic AI.

Daniel Hulme

Chief AI Officer, WPP

published on

15 November 2024

Colourful speech bubbles on a navy background

The term "agentic AI" has become a popular buzzword, with Google search volumes surging significantly since February. This growing interest highlights the eagerness of companies to leverage its perceived potential. Business leaders are finding themselves asking: What is agentic AI? How does agentic AI differ from traditional AI or generative AI? And crucially, why is agentic AI important for business?

However, amidst the excitement, there's a critical misunderstanding of what agentic AI truly is and how it can be harnessed effectively. This misunderstanding risks leading businesses down a path of misdirected investment and ultimately, missed opportunities.

Understanding agentic AI: beyond simple decision-making

AI research has been cyclic over the past 70 years, oscillating between single-agent systems and multi-agent reasoning. However, we are now transitioning towards a new phase where multiple AI agents can collaborate to generate new knowledge and better problem-solving capabilities. In a multi-agent system, different agents with specialised knowledge can work together. For example, if one agent knows that Socrates is a man, and another knows that all men are mortal, they can only deduce that Socrates is mortal through shared reasoning and communication.

One common misconception is equating agentic AI with systems that simply possess agency or the ability to make decisions. While decision-making is a component, true agentic AI goes far beyond that. It’s not just about AI making isolated choices; it's about creating agents with distinct perspectives and reasoning abilities that can collaborate to solve complex problems.

Another misunderstanding is viewing agentic AI as merely a collection of personas or audience simulations. While these can be useful, they don't capture the essence of agentic AI, which involves dynamic interaction and communication between agents to generate new insights and solutions.

WPP's approach to agentic AI solutions

At WPP, we're actively developing agentic AI solutions that embody these principles. Our Brand BrainsTM, which underpin our intelligent marketing operating system WPP Open, provide a prime example. Instead of relying on a single AI trained on vast amounts of data, we create a consortium of AI experts. Each expert focuses on a specific domain (e.g., tone of voice, imagery, written copy). This collaboration, mimicking how real-world experts work together, allows for the creation of higher quality and more nuanced advertising content.

We’re also developing AI Brains to represent various audiences, or even objective entities like food compliance or sustainability claims. This allows us to test content against these "representatives" to ensure it's not harmful or breaking any laws. This ethical and responsible approach is crucial when dealing with such powerful AI systems.

Avoiding pitfalls in agentic AI implementation

So, how can businesses avoid the pitfalls of chasing the "agentic" label without substance? Here are some key points to consider:

  • Focus on collaboration: Look beyond individual AI agents and prioritise systems that facilitate communication and knowledge exchange between them. This is where the true potential of agentic AI lies.
  • Invest in reasoning: Don't be swayed by systems that simply regurgitate data. Seek out AI that can reason, break down complex problems, and generate new insights.
  • Seek expertise: Building effective agentic AI requires deep expertise in areas like multi-agent systems and symbolic AI. Partnering with specialists or building internal capabilities is crucial. WPP, with its deep bench of talent across the network and within our AI research company, Satalia, is well-positioned to guide businesses in this endeavour.
  • Embrace the long game: Agentic AI is still in its nascent stages. Don't expect overnight miracles. Invest in research, experimentation and building a strong foundation for the future.

By shifting the focus from buzzwords to the core principles of agentic AI, businesses can unlock its true potential. This means moving beyond the hype and investing in systems that foster collaboration, reasoning, and ultimately, a more intelligent and effective way of solving complex problems. WPP is committed to being at the forefront of this AI revolution, empowering businesses to harness the power of agentic AI responsibly and effectively.

A version of this article was first published on Little Black Book.

Category

Technology & innovation Article

Related Topics

Artificial intelligence Industry insight AI

Explore More Topics

B2B Capitalising on creativity Commerce Health Talent

Show less

Explore More Topics

More in Technology & innovation

Panel of people sitting on a stage at an event at WPP's Sea Containers office

Are marketers prepared for the oncoming world of AI regulation?

As businesses look to maximise the possibilities of AI, how can they respond to the rapidly evolving regulatory landscape without limiting innovation or risking their reputations?

Cover shot of amplify: The Sound of Beauty magazine

From beats to beauty: how does the beauty sector sound?

Exploring the untapped power of sonic branding in the beauty world