Here's a quick rundown of the brilliantly organised and interesting ISBA AI Summit held today (03.10.2024) at Meta, bringing together marketers, platforms, advertising professionals and even a couple of Lewis Silkin lawyers.

The uses of AI in advertising and marketing have gone far beyond the early use cases of identifying audiences and, more recently, creation of content. It is adding to each step in the process. Below are some key points we picked up on:

Tech/Rollout

• Meta has invested big time in hardware, and the number of Ray-Ban x Meta glasses around the room and on screen would have won approval from ‘Mark’. The commitment to AI-powered hardware is evident.  
• AI is getting portable. Most AI tools need a good connection to a massive datacentre with microchips whirring away. Some low power AI is making its way onto devices, which could be transformative for use. 
• There are lots of tools not being made available in Europe (including the UK) because of regulatory uncertainty. 
• Trust in AI is broadly split between Western/democratic countries and less/undemocratic countries. Those in less/undemocratic countries are much more trusting of AI than those in democracies. The data shown on this actually drew gasps!
• Deploying AI in businesses is hard. The biggest challenge is getting data in some order and then showing people the benefits. 
• Building trust in AI is critical. Some ideas for doing so included increasing AI literacy, agreeing use/governance in contracts, conducting risk assessments, rolling out responsibly (bearing in mind the huge environmental impact of AI, and potential impact on the workforce), avoiding the more creepy use cases, such as AI-powered surveillance of consumers, and being transparent. 

Use in advertising

• Advertisers have gone from buying space, to buying audiences, and are now buying outcomes. AI is powering this more than ever. 
• AI is being used throughout the advertising production chain, from illumination (insights) to ideation and through to creation. 
• Brands are multimodal (i.e. visual, text based, audial etc.) and AI can help to analyse and produce multimodal content. The Meta ‘discovery engine’ (i.e. algorithm) prioritises creative content. 
• AI is enabling advertisers to see themes of their historic advertising (including the sorts of images, copy and even the tone/emotional impact)  to create new, on brand advertising.
• Advertising is moving from one hero idea/spot, and moving to a system of ideas. This not strictly an AI thing but AI-generated content that fits within a system of ideas for a campaign seems likely. 
• Advertising is becoming localised. New AI tools are allowing dubbing for local (possibly even regional) languages and accents, even modifying lips to match the words being spoken (not in Europe!).
• AI is aiding accessibility, with WPP, Microsoft and Unilever combining to produce simple and cheap audio description tools for use in campaigns. 
• Advertisers and agencies are implementing AI to make efficiency in the same way other industries are. Many advertisers referred to how much time their staff spend doing financial reporting, and the amount that can be sped up is mind-blowing. 
• Still only 52% of advertisers have a Generative AI policy, with 10% not knowing if they do. 80% of brands/advertisers are concerned about the use of AI in their supply chain. 
• There is a widespread recognition of the need for trust and transparency in the use of AI in advertising. People generally seemed to agree that where AI is being used it should be disclosed, but there is uncertainty over where to draw the line, e.g. when the AI use is minimal/background.

Regulation

• The UK has an opportunity to become a blueprint for sensible regulation, somewhere in between the position in the US and EU. Sounds good but is the UK a big enough market to set its own rules?
• Advertisers and agencies are more likely to be ‘deployers’ than ‘providers’ under the EU AI Act (a distinction that dictates the level of responsibilities organisations have, and is perhaps comparable to the controller-processor split in data protection law). However, where they team up with a provider, and put their own branding on it, they could be crossing the divide. They need to be aware they are doing so.