With the rise of AI, more people are asking ‘what are the benefits of a marketing agency when I can just use AI’? AI is being used increasingly in marketing and there is no denying that it is useful. But does it have the power to completely replace humans?

 

The drawbacks.

AI does have a few large drawbacks, of which most are centred around its lack of human personalisation. AI-generated content can be technically accurate and logical, but it lacks the human connection and emotional edge most consumers look for and most importantly, relate to. 

Take the 2024 Coke Christmas advert as an example. It was a fully AI-generated advert that left many people feeling disconnected or describing it as an “uncanny valley”.  Overall there was a generally sour attitude towards the advert.

 

The practical applications of AI. 

AI and Copywriting.

AI is great at developing ad copy, however, as it can only go so far it won’t be replacing copywriters any time soon. Now AI is able to create some great copy, yet humans are still needed to optimise this AI-created copy. After all, we are marketing to humans, not machines.

With AI, there is also the risk that as more marketing teams incorporate it into their work, more and more adverts will begin to all sound similar. Blending into one. AI-written copy is often very noticeable if not tweaked or developed by a human. One of many tell-tale signs is the presence of an Em-Dash (Which is bad news for the human copywriters who have used and loved them for years).

There is also the argument that because AI takes its information from other websites. Without human input, AI it loses all originality and authenticity (especially as we begin to see content created at scale by AI). So, whilst AI can help copywriters, it does not completely replace them. There will always be a need for human emotion as long as we are marketing to humans. One Stanford lecturer theorises that up to 95% of decision-making is done based on emotion. 

 

AI and data analysis.

It’s true AI is stronger on the data analysis front, but it can struggle to understand the full complexities of what people’s reasonings are behind their actions. 

Yes, AI is great for personalisation and audience targeting. It can help you tailor emails and even potential sales forecasting, allowing marketers easier access to their data than ever before. 

However, it takes a human to understand what external events (Such as valentines, birthdays or other events) may be driving your product’s purchases and connect with the consumers on an emotional level.

It is also important to remember that AI can still make errors, from having its own inbuilt bias towards topics. As AI learns from open source and human knowledge, it can easily pick up different types of bias from these data sources, and develop algorithmic bias. Many types of bias can influence AI’s decision-making, but Chapman university proposes five main types.

Five types of AI bias

Selection Bias 

This occurs when the training data does not accurately represent the real-world population, therefore making an incorrect assumption. If a facial recognition model is trained primarily on one skin tone, it may struggle to recognise those with other skin tones, leading to discrimination.

Confirmation Bias 

This happens when the AI relies to heavily on previous information it has learnt. This means it reaffirms previous historical bias, or bias that it has learnt before. Such as the stereotype that women are less qualified to work in tech. This alongside Stereotyping bias was a huge factor in causing the Amazon scandal.

Measurement Bias 

Similar to sample selection bias, measurement bias occurs when the data collection is not completed, or there is an error during collection. For example, if a university sends out  a course satisfaction survey, but only contacts those who stayed enrolled the full three years and not members who dropped out or changed courses, they’re not getting the full picture.

Stereotyping Bias 

Also called Prejudice bias. This is a bias that is trained into the machine and reflects societal prejudices, stereotypes and assumptions.  Many virtual assistants, such as Siri or Alexa, are designed with female voices and programmed to be polite and accommodating. This reinforces the stereotype that women should be helpful and submissive, which can help unintentionally perpetuate gender biases in society.

Out-Group Homogeneity Bias

This is where AI generalises data groups that it has less data on. Grouping together smaller groups even though they may have very few similarities. For example Facial recognition systems have difficulty distinguishing individuals from racial or ethnic minority groups. This is due to a lack of diversity in the training data.

 

Historically, these biases’ are also more likely to affect minority groups, so It’s essential to be aware of them. Take everything with a grain of salt, and be aware of contextual errors too. Most AI programs do not contain deep contextual understanding. Contextual errors can also happen due to AI’s lack of external knowledge. This is because if AI is not told about something it can’t consider it. For the most part, just keep in mind that you will often get out what you put in. So make sure you’re asking it quality questions.

 

So what does this all mean if you’re considering going 100% AI marketing?

Overall, AI is a great tool, but only when used with additional human input. Using AI to support your marketing work can increase productivity but not necessarily quality. 

Also, a word to the wise, don’t be mistaken – do not rely on AI to do the hard work for you, as it is not a replacement.  AI can streamline some of the work, but data (and content creation) still need human analysis and checking to make sure it has the authenticity and emotion to turn it into something worthy. So don’t worry, AI won’t be replacing marketers any time soon. 

 

To read more articled like this, head over to our blogs page.