Next AI wave involves solving challenges around machine consciousness: Daniel Hulme of WPP

New Delhi: Over the past two years, AI has led a significant disruption. AI today can generate images and videos, solve math problems, and even provide solutions for end-to-end logistical challenges. As per Daniel Hulme, Chief AI Officer at WPP, the next AI wave will involve solving challenges around machine consciousness. 

Hulme took to the stage on Wednesday at a day-long event named ‘Brew’ organised by GroupM in Gurugram to discuss machine consciousness, which is said to be the next wave of AI. 

Currently, AI possesses no consciousness and is dictated by mathematical equations called algorithms. “However, as a society, we do run the risk of potentially building conscious AIs and it is the duty of developers on society’s part to care for people and animals. Solving machine consciousness safely and responsibly will be the next big wave of AI,” added Hulme. 

Keeping in mind the goal of optimising end-to-end marketing, Hulme mentioned key challenges. “Firstly, most organisations are grappling with the issue of increasing and improving productivity. How do we write emails faster or how do we communicate more efficiently across our organisation are some of the challenges that employers are facing every day. Tools like Copilot and Duet can provide everybody with a productivity improvement.” 

The next challenge across the table involves optimising supply chains and removing friction points. Reaffirming his thoughts on the subject, Hulme said, “When it comes to supply chains, there are friction points that can be solved by AI. The challenge lies in identifying those friction points and creating a digital simulation of the supply chain like a digital twin to experiment with optimisation methods.”  

Citing an example for the same, Hulme said, “If an FMCG brand approaches a retailer and asks them, we are going to run a marketing campaign that will increase demand by 10%. Can you tell me if your suppliers will default on their supply? Do you have enough space in your warehouses? Do you have enough distribution drivers? Do you have enough people in your stores to fulfil that promise to the customer? All these questions can be answered using AI and brands have started to realise that they can use AI to solve those frictions.” 

To solve such major supply chain issues, WPP launched its own AI optimisation platform called WPP Open, which allows the agency to come up with out-of-the-box ideas and ideate them rapidly to create content and test that content against audiences at an unprecedented pace. 

Recently, enough brands have started employing large language models (LLMs) such as ChatGPT and Gemini to create and ideate content more rapidly. Stating a fundamental problem with using LLMs in the content ideation process, Hulme said, “While LLMs do allow us to create content rapidly, their results are generic, and generic information is not helpful as it is not differentiated. The battleground for the future is creating production-grade, brand-specific, differentiated content.” 

LLMs are also called brand brains. Hulme suggests four ways to make a brand's brain smarter. “Let's say you're trying to create a brain that creates brand-specific content like ads. By asking better questions, you can increase the sharpness of the brain by 20%.  

The second way of making it smart is by feeding it with brand guidelines, a specific tone of voice, and differentiated cultural values. Feeding contextual data will improve the quality of responses by LLMs up to about 40%.”  

Adding to the bandwagon of ideas enveloping the ability of AI to drive safer content consumption, Hulme said, “For the first time ever, we can show content to AI and gain feedback on how people think and feel about it. If I show you an ad, I don't know what was going on in your mind or body, but for the first time ever, we can recreate those signals and then use those signals to create better content and predict activation. AI models allow brands to create content and test it simultaneously.” 

For instance, if a person sees an advertisement with a black cat, a good marketer can predict the clicks, likes, and sales. But what machine learning can do is tell the brand that if they change that from a black cat to a ginger cat, the ad piece will get more clicks, likes, and sales because people relate a ginger cat to the comic character Garfield, bringing in relatability in the ad piece. No human being would have identified that complex correlation.  

Predicting the future of AI in the next ten years, Hulme said, “I think over the next 10 years we're going to see an explosion of new opportunities for people to contribute to humanity. One extreme is that AI can free up all jobs, which might lead to technological employment and social unrest. And we need to face that problem with policies like Universal Basic Income. The other extreme of the argument is that we should automate the production of food, education, healthcare, energy, and transport. If we can use AI to remove the friction, create those goods, and disseminate those goods, we can make everyone free so that people can go out and focus on what they always wanted to do now that AI is taking care of the laborious tasks.”