Are you curious about how chatbots might change how we work? Find out what ChatGPT is, and how it will impact the programmatic industry.
In part one (you can read it here) we looked at what ChatGPT is, how it can be used by marketers, and the way it may change how the programmatic landscape looks.
Now it’s time to flag the limitations, and share with you the things you should be cautious of when considering relying on ChatGPT for more than just weather updates and quickfire questions on ‘what to do in Lisbon this weekend’.
Here are four of the biggest weaknesses it’s important to be aware of when using ChatGPT – in particular regarding programmatic.
1) ChatGPT has a limited understanding of context
While ChatGPT can create human-like text based on the patterns it has learned, it doesn’t fully understand context. So it will sometimes return responses that are irrelevant, inappropriate, or even offensive. This is important to consider when generating text you may plan to use as content.
So while ChatGPT may be able to generate hundreds of programmatic ad headlines for you, you will still need to review and edit them before using.
2) ChatGPT can’t match your tone of voice
There’s far more to powerful communication than simply structuring a message. And one common complaint we’ve heard from people experimenting with ChatGPT is that, while its responses may be factually accurate, it hasn’t got the tone of voice right.
There’s also a certain ‘robotic-ness’ to its responses. Much like Spock from Star Trek, it might be able to convey information, but there’s no emotion behind it, and it can sound a little cold and flat – even when it’s using humour.
Again, this means you will still need a human being to check that any programmatic ad headlines and copy are on brand and in line with your tone of voice.
3) ChatGPT’s responses aren’t unique
Another reason why it is not wise to rely entirely on ChatGPT is that its responses aren’t unique. There are only so many ways to convey the same information, so if more than one person asks ChatGPT the same question, there’s a chance they will get the same reply – or that chunks of the response will be identical.
To test this we asked ChatGPT to create some content for us, then we checked that content to see if it existed online already. And it did – more than once.
So if you want unique text, which is important for programmatic ads, SEO and your reputation, it’s not safe to rely solely on ChatGPT. By all means use ChatGPT for inspiration and structure, but you’ll still need to check the veracity of its response, and put it into your own words if you want quality, trusted content.
4) ChatGPT isn’t always accurate
Like any AI model, ChatGPT is only as good as the data it is trained on. And if that training data contains biases or inaccuracies, these will be reflected in the responses it generates. So make sure you check any answers provided by ChatGPT before using them.
In general, like any tool, while ChatGPT has many impressive capabilities, it’s important to be aware of its limitations, and to use it appropriately.
ChatGPT won’t be the last big change to programmatic advertising and online content
Anyone who works in programmatic and content won’t be surprised to learn that ChatGPT will impact the way we act online, as the internet is a continually evolving environment.
And there’s no indication yet that chatbots will remove the need for humans when it comes to generating content for programmatic ads; instead it is anticipated that they will be tools that are used by them.
One thing will remain the same, whatever direction ChatGPT takes. And that is good quality, relevant information will always be important. As will staying abreast of any changes and acting accordingly. ChatGPT certainly won’t be the last new technology to impact how we behave online.
If you’d like to learn more about Chat GPT and its benefits, read our blog here and to find out how programmatic advertising can help you achieve outperforming marketing campaigns, please get in touch with one of our experts today.