If you are not paying for the AI products you use, you are the product

For strategic communication advisors, it is essential to adopt tools that increase efficiency, save time, and ultimately clients' money, but as a workforce often trusted with client information, there's also an obligation to move with caution, and remember there's no such thing as a free lunch – or a free AI tool.

There is a tenet that applies to all new technology: if you aren’t paying for the product, you are the product.

It took a while for users to grasp this with social media, but eventually they realised they paid for free platforms like Facebook and Instagram with their data, which the tech giants used to sell targeted advertising and make billions.

AI tools are no different, and users risk paying with not just their data, but also their skillsets, and possibly even their secrets. The nature of many generative AI tools and large language models (like ChatGPT and Google Bard) mean every interaction is a learning opportunity for the base model, and every back-and-forth with a chatbot is potentially another lesson taught.

Strategic communication advisors need to be particularly wary, not only because they might be training their replacement, but because they could accidentally divulge clients’ confidential information. Take a scenario: a consultant working up material for a carmaker’s new electric vehicle. The consultant might use an AI to streamline their research and compare the range of the new car to competitors’ vehicles. The consultant might then use the AI to draft promotional material. They might ask the AI to tweak the copy, bring a statement higher, or phrase it in a more casual tone. In one swoop, that consultant might have educated the AI about what features are most important in selling a car, how to write a good press release, and divulged sensitive information about the new car’s range capabilities.

Those learnings may not be incorporated into the model immediately, however as OpenAI states: “Your conversations may be reviewed by our AI trainers to improve our systems". Users can toggle their settings to stop conversations being used in this way, but how many people do is questionable. Add that to the black-box nature of these AI systems, and there may be cause for concern.

Some big players are already thinking this way. PwC banned employees from sharing firm or client data with ChatGPT in February, partly due to accuracy concerns over the content being produced, but also because of unresolved “intellectual property, security and privacy issues”. Amazon urged employees not to input any confidential information or code into the chatbot.

When I was a journalist, I interviewed some of Salesforce’s top executives in San Franscisco as the software giant unveiled a host of their own AI tools, and I put this exact carmaker scenario to them. The same word kept cropping up in their responses – “siloed”. If you paid for a Salesforce AI, I was repeatedly assured, the content, learning, and insights were retained. They were not shared with the wider model, or used to train it, reducing the risk of a competitor benefitting from the lessons the user imparted.

So it might be worth considering paying for the AI products used. This also insulates against another rude awakening coming down the line – bill shock. This has happened before with cloud provides. Many businesses were hooked by free or cheap data storage, and then got a nasty surprise when prices went up, or they received unexpected bills for additional space or tools. This left a bitter aftertaste and an unexpected expense on the balance sheets, and a similar situation is likely to play out with AI tools.

Right now in the tech world, it's all about being first, and gobbling up market share, and that's why so much is being given away for free. That won't last forever. In the long run, it may be best to know the price-of-use now, rather than finding out the true cost of using these "free" AIs later.

At Baldwin Boyle Group we are developing policies that enable us to harness the benefits of AI for us and our clients while also protecting intellectual property, confidentiality and workers’ dignity. Our business is based on rigorous analysis, insight, intellect and imagination and for now, at least, these remain the domain of humans.