What I’ve Learned About Successful AI Implementation – A CTO’s Perspective
In the past year or so, AI-powered tools have delivered increasing value for people and businesses. Not only is the technology getting better, but people are also better at using it. Collectively, we’re starting to understand what AI can do well and what its limitations are.
Many organisations are already on the road to AI implementation; but their journey has only just begun. A year on from my previous blog on the secrets to successful AI implementation, those key principles still hold true. In fact, you could argue that some of them – such as designating a ‘data steward’ – have risen in importance.
So what’s changed in the last year or so, and what do we now know that we weren’t certain about before?
Here’s what I’ve learned about successful AI implementation.
By Martin Nürnberg Gundertofte, CTO at WorkPoint
It’s vital to validate and verify AI-generated content
When AI tools like ChatGPT and Stable Diffusion first became publicly available in 2022, there was a huge wave of excitement and optimism about the potential of AI. Time has shown that there’s a learning curve as people get to grips with AI’s abilities and limitations.
We’ve come a long way from AI chatbots falsely asserting that ‘Leonardo da Vinci painted the Mona Lisa in 1815’. Since the early days, both the technology and the people using it have steadily improved. But what those early lessons have taught us is the golden rule of creating content using GenAI – always validate and verify what you see.
Fact-checking might have fallen out of fashion amongst social media platforms, but for most organizations in both the public and private sectors, it’s a mission-critical imperative. Incorrect information erodes public trust, tarnishes your brand image, and ultimately hurts your business.
Outside the workplace, wider experience and experimentation with AI-powered tools, both at home and in education, means people are becoming better educated about how to use, and how not to use, AI. With that learning comes the understanding that validation and verification are key to unlocking the true value of AI.
Large Language Models are improving all the time
Not only are people getting better at using AI tools, the Large Language Models (LLM) behind AI-powered chatbots and assistants have also massively improved. They still have quirks and produce anomalies – validation and verification are essential – but with strong feedback mechanisms in place, the quality of their output is steadily increasing.
From a business perspective, it’s important that people know how to validate and verify information that could ultimately lead to a decision based on wrong information. The encouraging thing is that AI-generated content – and our ability to refine and rework the output – means the quality of output from today’s AI tools is arguably much higher than even a year ago.
Choosing the right moment to introduce GenAI tools into your business is a decision for each individual organization. Platforms like WorkPoint365 can help you become AI-ready, but knowing how and when to implement it depends on your business model. In many cases, it will take more time for AI-powered tools to become comfortably and reliably integrated into everyday workflows.
AI tools are slowly winning people’s trust
The clue is in the title. Large Learning Models have proven to be very effective learning tools. You only need to scan the top of a Search Engine Results Page (SERP) to see the faith we now place in the reliability of AI-generated content.
As AI models continuously improve, human trust in their ability to deliver the right answers – with careful verification and validation – increases. From researching a new topic to content creation – whether it’s documents, videos, images, or music – AI models are capable of producing impressive content, albeit moderated and modified by human oversight.
The dawn of Copilot in the Microsoft ecosystem, and similar implementations from the other big techs, means AI is becoming an integral part of our daily lives. From helping draft emails to gathering information from spreadsheets, these small enhancements give us the edge and facilitate small efficiencies that start to add up when they’re compounded across an entire organization.
So what’s next for AI tools? Discover more in my next blog. Coming soon.