Responsible AI
We look behind the veil at ethics and responsible AI.
I’m sorta thrilled to say that I’m setting the agenda for this one. In a former life when I was a very bald, very male man in black and white, I suggested we should do a piece on ethical AI. My wish is my own command… yes, it’s me,
, taking it from the top this week.AI is not some all-knowing, all-fair, perfectly neutral genius, it’s more like Joey from Friends taking a ‘How Well Do You Know London?’ quiz after one day in the city. It’s only as good as the data it’s been trained on, and that data? - flawed. That’s why keeping a ‘human in the loop’ is absolutely essential. AI makes decisions, but humans need to check, guide, and correct them because otherwise, we risk making very bad choices at very high speeds.
Why is AI Biased?
Think of AI like a scientist explaining something. It might sound convincing, but that doesn’t mean it’s correct. AI learns from past data, and past data comes from us (the developers), and let’s be honest we’re a little messy. AI absorbs patterns, including the good, the bad, and the wildly problematic. If historical hiring data favoured certain demographics, AI might do the same. If social media conversations reflect stereotypes, AI will pick them up. It doesn’t mean to, but it doesn’t know better unless we teach it.
Bias to think about
Representation Bias - if the training data mostly represents one group, AI will struggle to serve others fairly.
Confirmation Bias - AI can reinforce existing patterns, even if they’re flawed.
Automation Bias - the idea that if AI says it, it must be right! No, no, sometimes AI is just confidently wrong.
Societal Bias – if AI is trained on biased human decisions, it repeats those biases.
How Do We Keep AI Ethical?
Human in the loop. AI shouldn’t make final decisions on hiring, legal cases, or medical diagnoses without a human reviewing the outcome. You are the last filter.
Diversify Training Data. AI needs exposure to a wide range of perspectives, cultures, and experiences to avoid defaulting to the same old biases.
Regularly Audit AI Systems. AI needs constant monitoring to make sure it’s behaving fairly.
Transparency is key. We need to know how AI reaches decisions. If it’s making choices based on flawed logic, we should be able to step in and fix them.
Challenge the Output. AI is like a well-meaning but occasionally clueless friend, so question it. If something feels off, double-check the sources and reasoning before acting on it.
AI isn’t good or evil, it’s just a reflection of us, which means it carries all of our biases, flaws, and occasional irrational decisions. The key to making AI ethical is humans. We have to guide it, correct it, and never assume it’s automatically right.
So let’s make sure AI isn’t misunderstood, misrepresented, and in need of a human to step in and set things right.
Dettman Reflects
I have noticed that AI does bring out degrees of fear in people. It goes beyond just the normal reluctance to engage with technology or new products. People genuinely misunderstand its ways and I think a lot of this is down to the name. It most certainly is artificial but it is a long way short of intelligence. This might help.
Gartner have some memorable sayings. One of the most relevant is this one: there is no Artificial Intelligence without Human Intelligence. It works on many levels.
My advice, and it’s not just me saying this, as you now realise, is to forget over-thinking about the AI label. It’s just a bunch of stuff that can help you at work. Let others worry about all those ethical things. As long as you are being ethical and true to yourself in the ways you use the tools then you have nothing to worry about. If people like Joey want to create images of unimaginable scenes then leave him to it.
Big Tech will always prioritize progress and performance over principles. [Many Ps. Danni.] You just have to trust the industry and governments and public perception and pressure will get us to the right place in the end. [Way too many Ps now. Danni.]