If you’re a marketer, you have likely seen a lot around AI recently. ChatGPT has massively accelerated the use of Generative AI. It reached 100m users in 2 months. By comparison, it took TikTok 9 months and Instagram 2.5 years to get to the same level. -reference
Sonar has been experimenting with Generative AI in a variety of ways. Mainly, ChatGPT to help create user flow, test scenarios, UX/UI mapping and researching. We also specifically use it as a chrome extension alongside Google for SEO. MidJourney and Dall-E2 for website and social media. We are trialing using AI within digital performance campaigns and as part of keyword research.
We have also used ChatGPT for ‘fun’ – naming our fish, using it to come up with social ideas for the team, etc.
Digital marketers have found some serious use cases of AI to save a considerable amount of time, and enhance the end product.
Some examples:
- Asset creation: Creating content that is both high-quality and unique. For example, creating bespoke imagery for social media that is still in line with the client style guide.
- Scaling Content: Creating content at scale through generating multiple versions of a marketing message with slight variations. For example, trialling multiple messages and formats within a performance campaign to see which variant was most effective.
- Dynamic Content Creation: Generating content that is more likely to be engaging and relevant to users (by analysing data on user behaviour, preferences and interests). For example, feeding in a number of images and correlating with user engagement data to deduce which content is more likely to (a) engage, vs (trigger a sale), etc.
- Enhance personalisation and product recommendations: Similar to above, if we can combine user engagement and product information we can combine data sources to suggest what product might be the best match for the user.
But – it also has a number of drawbacks, largely based on the fact that it creates new data based on existing data.
- It can create false or misleading information. In this case, it can be a matter of ‘garbage in, garbage out’. If the data being used is inaccurate, the output of generative AI will also be poor, misleading, or false.
- It can be biased. An example is when there was a prompt of Generative AI to show doctors. The output was almost entirely white males.
- It can be vanilla. There are a variety of posts from copywriters illustrating how Generative AI results are quite bland and ineffective relative to copywriters with more of a grasp of the spoken language.
While it is very early days, it is clear that Generative AI is now a critical component in the digital marketers toolkit – and is a ‘partner’ in any discipline relating to digital marketing. As a Sonar team member said – “I don’t think Generative AI will be taking our jobs, but digital marketers will need to know how to use it to keep our jobs”.