AI: DIY, or off-the-shelf?
When it comes to incorporating artificial intelligence, the fundamental question for businesses is, as with so many other similar decisions, whether to use the algorithms and models sold by technology companies such as OpenAI, Google, Meta, Anthropic and others, or instead try to develop their own, a task erroneously seen by many as unmanageable and very complex.
Initially, companies began testing algorithms like ChatGPT because they were easily available. The rush to find out if ChatGPT knew anything about them, their products or managers was followed by a justified wave of fear when it became clear that the use of such algorithms did not guarantee the privacy of any information that had been entered. In fact, the information they had given was being used to train these algorithms, and could even be used later, as growing numbers of secrecy-obsessed managers soon began to find out.
We should remember that generic algorithms are trained with information that has been collected and selected, generally with little filtering, from the internet. If your processes, products or the information related to them are not available on the internet, chances are that a shop-bought algorithm will likely provide generic or simplistic answers as soon as you try to dig a little deeper. The same algorithm that writes great messages, translates wonderfully, makes a presentation or…