What is known as “AI” and/or “machine learning” is becoming more prevalent in the work environment. From the “bad algorithm” upset GCSE Grading until Microsoft 365 finds itself working with just one document all day, there’s no getting away from the fact that there are more “automated” services than ever before.
For DPOs, records managers and IG officers, this poses some interesting challenges for the future of records, information and personal data.
At a recent IRMS Public Sector Group webinar, I was asked to speak on the challenges of AI and machine learning. In a session titled “AI and his IRM Challenges of what is called the ‘Metaverse’,” he examined a variety of issues. I would like to briefly touch on some of them below. I’m not convinced that the “metaverse” will happen anytime soon, but AI and algorithms are right here and growing fast.
From a personal data and privacy standpoint, we know algorithms guide our online lives. From the ads we see to the posts we see in feeds like Twitter and Facebook, it’s no surprise that this algorithm knows me better than I do. And how will it go before it has a massive impact on everyday citizens? What if a “social algorithm” resolved your sexuality before you and your family? I still work with families who abuse and kick out LGBTQ+ people.
An interesting question was posted on Twitter about deepfake videos and whether they are personal data; The answers are mixed and have interesting implications for the future. If someone could use AI to generate a video of you doing something you didn’t mean to do, can you imagine the impact? By then, the damage will occur.
In an organizational context, the simple logic is that if an organization uses an algorithm to aid workloads and efficient services, the algorithm must be perfect. As the Borg Queen (cybernetic alien) from Star Trek First Contact once said to Data (a self-aware android), “You are an imperfect being created by imperfect beings. Find your weaknesses. It’s just a matter of time.” If anyone finds a perfectly designed system without process issues, bugs and reliability issues, please let me know.
Many data scientists and leading data commentators such as Cathy O’Neill frequently state that “algorithms are basically opinions embedded in code.” And opinions bring room for bias, fault, and error.
That’s not to say these things don’t have their benefits, though. There are so many advantages. However, to get anything good out of them, you have to make sure good things go into them and good things help create them. Supplying them with a constant supply of power always leads to problems, as we’ve seen time and time again.
myself and Ol I was present (in effect, Olu in person) at the launch of AI standard hub By Allan Turning Institute (and others). An attractive initiative by ATI and others, including the UK Government.
Now why am I talking about this? Simply put, as I said above, this technology is here and it’s not going anywhere. See this company’s offer live editing of your voice and you may find this conversation between Google engineers and AI very thought-provoking and sometimes terrifying. If anything, AI is evolving faster and faster. Therefore, information professionals in all fields need to be aware of how it works, how it can be used within an organization, and how they can upskill to challenge and support it.
Recently, ICOs have published various guidance on this. Relatively detailed guide Learn how to use AI and consider its impact on data protection. This is not a user manual, but it contains important points to consider and steps to take.
Yes, Alexa, quit my blog! Oh, I mean Hey Siri, quit my blog… Damn it… OK Google…
