By Jonny Bentwood, Global Head of Data & Analytics
There is a great deal of hype around AI within a marketing framework. Whilst the technology could be incredibly useful and adoption could drive efficiencies and competitive advantage on a global stage, there is currently a great delta between what it can do, and what it currently is being used for.
To start with, let’s distinguish between AI and machine learning – often used interchangeably and incorrectly. In summary, while machine learning is about understanding the past, AI is about working on the future. However, even with that understanding, there are far more misunderstandings around AI and what it can and can’t do.
AI taking over the world
There is a huge amount of incorrect fear about AI. This isn’t that the fear isn’t justified but rather that it is misplaced. AI does have the potential to be destructive as Stephen Hawking and Elon Musk have said- but most people when thinking about the dangers of AI think of Terminator.
In reality, we should think of AI more like the sorcerer’s apprentice in Fantasia. The danger of AI is not self-awareness, but our own inability to correctly frame the limitations of the question that AI supports. In Fantasia, the sorcerer asks his apprentice Mickey to fill the cauldron. Mickey, being tired, created a spell to get the broomsticks to do it for him. The problem is that he didn’t tell them when to stop. This is AI. Mickey teaching the broomsticks what to do but failing to add the limitations. Asking an AI to identify the best way to cure cancer (without proper limitations) could result in a diagnosis of human extinction.
Fortunately, AI is far away from this paradigm-shift in the agency-world, but the concept of AI being a quick way to answer questions, provided the framework is watertight, is crucial. We should be careful of our AI overlord’s support in case the person asking the question didn’t ask it correctly – without these checks and measures, our love of speed in an agency could result in poor advice to clients.
AI delivering little more than sentiment and relevance ‘currently’
The hype of AI is that it can solve problems in creative ways not imagined by ‘inferior human brains’ at a quick speed. Imagine the possibilities of coming up with a creative idea that is so radical that Cannes Lions will be up for the taking. This is the promise. Data and creativity without being limited to previous boundaries regarding what is feasible.
However, if this is the dream. The reality is far more mundane. AI is strong when it comes to:
- Sentiment analysis. Whereas sentiment coding has been problematic at best, current AI can understand attitude on an entity level and learn and improve all the time. This is available now and PR agencies are using this tech on a daily basis.
- Relevance. Consider Apple. Imagine how difficult it is to understand what is happening in their world when their name could easily be confused with a fruit, a cake or indeed a record company. AI has found a solid use case in understanding context and removing the need for humans to spend unnecessary countless hours categorising and coding.
The future: Value-based Pricing due to Scenario Testing
In 2021, I anticipate AI helping more when it comes to scenario testing. Why should an agency spend $$$ of our client’s money without knowing their idea will work. Combining the beauty of big data with the capabilities of AI-powered tech such as Cortana or Watson will enable firms to value price their work, as they will know the likely impact of their work before commencing. Work will be done quicker but with higher investment in tech – clients will have to pay for this helping hand to turn the model from hour based work to value-based.
Golin’s work on customer experience could be further enhanced as the AI understands the content that is being directed at the company. Instead of relying on complex workflow and rule-based process, the AI could deliver on-trend, identify opportunities and deliver on micro-changes that can result in huge impact
The Far Future
Already “AIs” are composing music and writing novels that the best Turing Test finds difficult to distinguish from human creations. Will this lead to fewer humans in the workforce if the AI can plan, create copy and implement the campaign. I doubt it. What will change is speed – for that AI will be needed. But for the rest – we will still need us carbon-based workhorses. If for no other reason, than I hope to keep my job!