Friend of FAIRER interview: Malik James-Williams

About Malik James-Williams

Malik is Head of AI Operations with more than 11 years' experience spanning media, marketing technology and artificial intelligence. Throughout his career, he has specialised in bridging the gap between emerging technology and the people who use it – making AI practical and accessible in everyday work. His work focuses on designing AI operating systems, workflows and frameworks that help organisations adopt AI responsibly, while ensuring outputs align with business objectives and quality standards.

AI is changing organisations, not just jobs

Malik highlights two significant organisational shifts that are emerging as a result of AI.

"I think AI is changing two main things within organisations, and they're happening at different stages. The obvious gain is productivity. You're able to ask a tool, such as ChatGPT, to help structure an email, enter data into a spreadsheet or produce a first draft. I don't want to dismiss how useful that is because, for some people, it genuinely saves hours of time. But I think it mostly makes individuals faster inside their existing processes. The organisation itself doesn't change very much.

"The second shift is the more interesting but harder shift – the move towards organisational redesign. This is where businesses build a structure – or an AI operating system – which evaluates artificial intelligence effectively, with a human in the loop, while also building a framework around the AI so that it can navigate your business properly. This means moving beyond just giving people tools and asking them to use them. A lot of organisations are essentially buying access to AI and calling it a strategy, but that’s not a strategy.

"For HR leaders, the shift is about redesigning how work actually flows. Do you still need someone doing repetitive data entry, or can AI take that on while people move into judgement-based work? Where does human decision-making add the most value, and how do roles evolve when the routine parts of jobs are automated?"

AI should augment people, not replace them

According to Malik, AI is most effective when used alongside people as a thinking partner, not as a substitute for human workers.

"One thing I always try to be clear about is that AI is strongest at execution, while humans are still best at judgement. Whether it's a presentation, task, report or an email, AI can produce a really strong first draft. The human is then the extra layer that uses judgement and expertise to tailor that work to exactly what the client, customer or business needs. We shouldn't be talking about replacing people with AI. We should be talking about augmenting people – speeding them up and freeing them to do more judgement work rather than execution work.

"Another powerful use case is treating AI as a thinking partner rather than simply an answer machine."

"Instead of asking it to complete a task, ask it to challenge your thinking."

"Build systems that ask follow-up questions, test your assumptions and anticipate the objections your manager, client or stakeholder might raise before an idea ever reaches them. That's a very different way of thinking about AI. Rather than replacing judgement, it helps people apply more judgement to their own work before another human ever has to review it."

How AI is reshaping early career development

Malik reflects on how AI is reshaping early career development and why clear communication and sound judgement will become increasingly important in the workplace.

"I think AI will reshape processes before it reshapes job titles. Anything that involves taking unstructured information, turning it into a finished output or checking that output against a standard is going to change. Initially, I think it affects the production layer first – people doing first drafts, formatting, research and collation. Traditionally, that's meant junior people.

"You do the basic work, someone more senior guides you and, over time, you absorb their judgement. You learn what a good output looks like by producing lots of imperfect outputs and understanding why they weren't quite right.

"My current view is that junior employees may increasingly start as AI reviewers rather than producers. That could mean checking AI outputs against a standard, comparing different options, spotting mistakes and learning what good looks like by reviewing a much higher volume of work than they would ever have produced themselves. That could accelerate learning, but it also carries a risk that people learn to recognise quality without getting enough practice creating it themselves. I think businesses will need to invest in junior people for longer, helping them build the judgement that ultimately allows them to move into more senior roles."

"One of the most important skills is specification – being able to clearly explain what you want from AI."

"A lot of senior people are used to giving vague instructions because humans naturally fill in the gaps. Machines don't. They simply give you back that same level of vagueness. The organisations that get the most from AI won't necessarily be the ones with the most advanced technology. They'll be the ones that can describe, with real precision, what a good output looks like."

While AI has the potential to improve efficiency, Malik warns that organisations must take proactive steps to prevent bias from being replicated at scale.

"One of the biggest risks with AI is assuming that it's neutral. Left on its own, AI will likely scale bias rather than remove it. Every large language model (LLM) is trained on data written by people, and people have biases, which means the outputs of those models are inherently biased as well. If we're using AI in hiring, promotion or decision-making, we have to recognise that we're not starting from a blank slate. A human recruiter might make biased decisions inconsistently, but a system can apply that same bias consistently across thousands of candidates. That's why the scale of the problem becomes so much bigger.

"A well-known example is an experimental CV screening tool that learned from historical hiring data and ended up penalising CVs that included references associated with women. Had that gone into full production, it would simply have kept hiring the same kinds of people the organisation had hired before.

"That's why I think reducing bias has to work from both directions. There's the top-down approach, where leaders set the vision and define what they want AI to do. Then there's the bottom-up approach, where you involve people from across the organisation in building those systems.

"If I'm creating an AI that writes client emails, I speak to client services. I speak to different levels of seniority. I speak to analytics, marketing and PR. I want to understand what good looks like from different perspectives before I ask AI to reproduce it. The people building those systems need to be diverse because that's how you reduce some of the bias before it becomes embedded in the technology.

"It's also important to show AI not only what good looks like, but what bad looks like as well. You need examples of both, so the model can reason about the difference rather than simply repeat patterns. Ultimately, humans should always be the final stage in the process."

Businesses need to decide what responsible AI looks like

Malik argues that fairness and accountability must be built into AI from the start, with clear ownership and ethical oversight to guide how the system is used.

"I think organisations need to give fairness an owner. At the moment, we spend a lot of time thinking about implementing AI, but much less time thinking about how we use it ethically. Some of the frontier AI companies are already moving in this direction. They're hiring people from completely different disciplines to challenge how these models behave and to think about the risks they create. I think organisations will need something similar.

"Whether it's a Chief AI Ethics Officer or another leadership role, somebody needs to be responsible for defining what fairness looks like, how bias is tested and how ethical principles are built into the foundations of an AI operating system. If you're not yet at the stage of building sophisticated AI systems, the same principle still applies.

"Even if you're using something as simple as ChatGPT, you should be asking how you use it fairly, how you reduce bias in the outputs and how you make sure people are treated consistently. Fairness shouldn't be something that's checked at the end; it has to be designed into the system from the beginning.

"At the moment, AI remains largely unregulated, and many organisations are still developing their own approaches to governance. That makes internal accountability even more important. Until stronger external standards emerge, businesses need to decide for themselves what responsible AI looks like."

Everyone should have access to AI, or you risk perpetuating existing inequities

Malik considers the risk of unequal access to AI tools and training in the workplace, and what organisations can do to ensure the benefits of AI are shared more widely.

"People from lower-income backgrounds, disabled people, people with lower digital confidence, people whose first language isn't English and groups that are already underrepresented in higher-paid knowledge work are all more likely to miss out if AI adoption is simply left to the market.

"Often, it isn't about whether someone can access ChatGPT at home, it's about whether they're allowed to use AI at work, whether they're given the right tools and whether they're trained to use them properly. Meanwhile, people already in higher-paying, well-supported jobs get better technology, more permission to experiment and more opportunities to turn AI into a career advantage. That's probably the gap I worry about most.

"Communication is another area where AI can make work more accessible. Whether English isn't your first language, you're early in your career or you simply lack confidence writing to senior leaders, AI can help structure your ideas and make workplace communication feel more accessible. But I wrestle with that as well. Ideally, people should be able to be themselves at work rather than feeling they have to write like everyone else. Used carelessly, AI could end up encouraging everyone to adopt the same corporate voice instead of valuing different communication styles."

"I think employers need to treat AI access and AI training as part of workforce development, not as a perk for people who are already senior or already confident with technology. Training should be available to everyone – junior staff, admin teams, frontline managers and support functions."

"If AI is becoming part of the basic equipment of work, then access to it should be funded and supported in exactly the same way as any other workplace training."

A word from Laura

Laura Drakeford, Inclusion Partner at FAIRER, shares her key takeaways from Malik's interview.

What stands out is how quickly AI can move from a productivity tool to something that can reshape fairness. Organisations can’t treat it as just another system – it needs ownership, guardrails and critical thinking applied, otherwise we risk perpetuating bias at an unprecedented scale. Leaders need to question who benefits, who gets left behind, and whether AI is amplifying existing inequalities or genuinely helping people work and communicate more fairly.

If you would like to be considered for an expert interview, or want to further discuss any of the themes covered in Malik's interview, please get in touch.