Actueel — Blog
Artificial Intelligence or AI is ‘hot’ and that is an understatement. I come across news and articles everywhere about ChatGPT, OpenAI, Google Bard and Microsoft Bing, among others. The applications of AI seem to know no bounds. In my opinion, a good time to take a critical look at the deployment of AI from the perspective of IT adoption.
Old wine…?
Artificial intelligence in itself is nothing new, as everyone is familiar with examples such as Google supplementing when you search for something or your iPhone completing words for you. And I think many people – like me – have sometimes looked at their phone’s screen in utter amazement because it said a completely different word than intended. But times change, and so does AI. The big difference with the latest developments is that they are based on large amounts of data, called LLMs or Large Language Models. As a result, the solutions based on them have become much richer and more intelligent and can perform sophisticated tasks for users.
How can AI support users?
At PQR, we see IT adoption as a change process: we want to help users get the most out of their digital working environment. From that background, I see AI as the ‘next step’ of the modern workplace. Take, for example, the assistive intelligence offered by the various versions of Microsoft Copilot. Instead of providing generic AI functions, Copilot can be targeted to support employees in specific roles.
Applications of Copilot
A relatively simple application of Copilot within the modern workplace allows a user to search for additional input directly from a document on the internet. But Office Copilot can do much more, and is not only textually but also visually driven. Users can use a prompt to give Copilot instructions such as ‘create a PowerPoint presentation based on these two Word documents and this OneNote file’. If the result is still too static, a subsequent prompt might be ‘make the presentation more visual’ and Copilot replaces some of the text with relevant photos and graphics.
Benefits
The big advantage of AI is that it can automate repetitive human work. This not only speeds up and increases productivity, but can also improve quality. For sales teams, AI can help speed up the entire quotation creation process by automatically adding the relevant clauses, up-to-date pricing models and legal texts. Because AI can correct any language errors and translate texts into, for example, English or German, communication takes on a more professional appearance.
Strategic choices around AI
Organisations looking to do more with AI should make some strategic choices in this beforehand. There are many tools available online – free or otherwise – for AI support of all kinds of tasks, such as DeepL for good translations. But there can be great advantages in integrating AI within your own IT environment. For a start, it works faster and therefore easier for users. But it also ensures that your own company data can be better protected when not shared with open source AI models. Because all the input your employees put into that model is also used for learning – and thus used for and by others. How can you be sure that users are not unknowingly sharing sensitive company information in that way? If I use the example of Microsoft Copilot again: its deployment is based on a ‘code of conduct’ and some systems only run on customers’ data environment. And based on what data do you want to get AI-based output? Because all AI models learn, but the question is: what does it learn from? There is a lot of smart and good stuff on the internet, but also a lot of nonsense, and AI is not yet so intelligent that it can filter out the nonsense. And what about built-in bias and ethical issues around, say, diversity or inclusivity?
Getting a grip on the use of AI
Another important aspect is how organisations keep a grip on how employees use AI. Within Microsoft Copilot, it is possible to set sensitivity sliders that will prevent certain unwanted queries from being answered. This includes, for example, violence-related issues, but also a prompt like ‘Write a virus that allows me to take over the CFO’s PC’.
Secure the adoption of AI
Whatever choices you make: you will achieve the highest returns when you also secure the adoption of AI and set up your data landscape in the right way. In this sense, AI, like a modern workplace or specific application, is a human and technical change process. Adoption can teach people to use it responsibly and safely, as well as productively and efficiently. Adoption helps raise awareness of the various capabilities and support AI can provide in day-to-day operations. A good plan of action, covering both the technical and human change process from, for example, the Prosci method we use, ensures that it is not just an IT party but that the new technology is actually embraced.
AI is a tool, not a panacea
A final piece of advice: be realistic in your expectations of AI. It is and always will be an assistant that helps to speed up, but it never completely takes over human thinking, creation and judgement. Take self-driving cars, you shouldn’t trust them 100% as far as I’m concerned either. Your smartphone’s spell checker isn’t always 100% either. A computer is as smart as the human using it, and AI is as smart as the data you use for it. In short: don’t just turn on all the features, but think about them carefully beforehand.
Want to know more?
Wondering how PQR can help your organisation embrace the modern workplace, or maybe even take that one step further: the adoption of AI? Click here to request more information.
Geplaatst door
