Generative AI and the evolution of search

MBM’s Chief Digital Officer, Emily Isle, along with Senior Search Director, Charlotte Lobo, delve into the realm of AI’s actualities, emphasising the need for continuous learning and offering a practical three-step approach to managing AI fever that keeps search specialists at the forefront.

Executive Summary

AI fever is taking over news headlines, infiltrating every industry, sparking imagination and alarm across the globe. But most thought pieces cover the same extreme ground – some on the side of AI enabling everything, others warning it’s the end of the world.

The reality is the fever has just heated up, and we all have a lot to learn. Advances in automation technology abound, but the entrancing promise of a surge in productivity with the minimal involvement of actual human workers is a farce. Productive automation needs the machine, the system, and the team to work.

Charlotte Lobo.

We all now have co-pilots in AI for our marketing endeavours. It is crucial we use this new tech responsibly and integrate it into our team dynamic. It promises advanced campaign optimization and content management opportunities. Where we see immediate impact is in the search marketing discipline.

So, with this in mind, we’re here to manage the fever in three, digestible steps that keep happy search specialists at the forefront: make it safe, listen to our team, and test, learn and repeat.

Make it safe

We’re familiar with machine learning algorithms like RankBrain, multitask unified models, AI writing tools and even large language models built into Google technology. However, with the Generative AI genie out in the open, what is new is mass access and engagement beyond in-platform application. 

When we test in-platform, everything is managed within ad tech’s data security. Now we have interaction in open interfaces, and the blend of AI-generated and advertising content in the same media experience. This is both exciting and worrying and poses important questions on brand safety and IP security.

Emily Isle.

Brand reputation, data protection and content suitability need to be addressed before ads on chat become a reality. The Italian Garante’s recent investigation into OpenAI’s data breach and lack of age verification to protect younger users from inappropriate generative AI content during registration is just the tip of the safety and content suitability iceberg. What is clear is that brand suitability technology is more important than ever. Tools like IAS, Zefr and Publicis’s proprietary Lighthouse are stepping up to the plate. These can be adjusted to a brand’s nuances and used widely across digital platforms.

IP security is also a big watch-out. As soon as ChatGPT hit mainstream, we needed to set guardrails on not asking questions inclusive of IP. We forged a partnership with OpenAI from 2017, and Publicis group released SandboxGPT in May 2023 – a proprietary interface for exclusive access to prompt-based Large Language Model (LLM) platforms in a safe way. These accesses protect our clients, protect IP, and ensure that our prompts and data will not be used to train the LLM platforms. SandboxGPT launched with ChatGPT and DALL-E basic, and Bard and Firefly are tee’d up to onboard soon.

Time to Listen to the Team

Our search specialists are consistently at the forefront of new technologies as they integrate seamlessly into search platforms. 

If you think the latest generative genie could replace Google Search, think again! “Saying ChatGPT will replace search is like saying podcasts will replace universities. They do two different things,” says Tristan Greene, TNW Author.

Rather than a replacement, chatbot technology will be a supplement to Search, integrated into the traditional search ecosystem. 

For search marketing, this isn’t the end of an era but rather the beginning of a smarter, context-aware, personalized model that requires new, innovative tactics.

We are revisiting SEO content strategy with Google’s ‘E-E-A-T’ (experience, expertise, authoritativeness, and trustworthiness) as a priority point of difference within the explosion of generative AI content. We are leveraging predictive SEO and Answer Engine Optimisation techniques. We also know that as generative search emerges, structured data will grow in importance, as will keyword and entity optimisation. 

Test, Learn, Repeat

How to get the best output? Focus on quality input – easy.

This is where team expertise comes in managing the output quality. We need to ask specific questions, read up on recommended prompt lists, give the bot as much context as possible and avoid jargon or slang. 

Once we’ve honed our questioning technique, we turn to experts in the industry for even more ideas – from search copy writing to competitive research to consumer behaviour insights. Our SEO professionals can also use the tool in a range of ways but as always, with caution, applying personal expertise, alongside regular SEO tools & strategies.

Whether it’s keyword research or ad copy writing, an expert needs to own the output. Silly as it seems, we don’t use generative AI to answer a question we can’t already verify.

Closing Thoughts

Generative AI changes the game, but we are here to win it. The relationship we have with generative AI isn’t a competition, it’s a partnership between the machine, the system and the team using the tool.

Technology is not a substitute for common sense. When used cautiously, strategically and for the right reasons, the results are generally commendable. Something our Search practitioners know all too well.

Comments are closed.