As artificial intelligence reshapes industries, HR leaders face the challenge of integrating it into their organisations effectively and safely. Drawing on insights from AI entrepreneur and academic Daniel Hulm, Siân Harrington explores 10 crucial lessons for HR professionals
“Companies don't have insight problems. They have decision problems.” And it turns out that humans are rubbish at making decisions. So says Daniel Hulm, entrepreneur in residence at University College London and chief AI officer at advertising giant WPP who has 25 years’ experience in artificial intelligence.
He adds: “What you want to do is build systems that make decisions, learn about whether those decisions are good or bad and adapt themselves so next time they make better decisions. If I held what we currently do in industry to this definition one might argue that nobody's doing AI.”
Hulm was a keynote speaker at the CIPD’s Festival of Work in London and his presentation certainly got HR professionals thinking. The People Space’s editorial director Siân Harrington pulls out key lessons:
1. Understand the core problem: decision-making over insights
One of the fundamental points highlighted by Hulm is the misidentification of core issues within organisations. Many companies focus on data collection and generating insights, assuming that better insights will naturally lead to better decisions. However, the real challenge lies in decision-making itself. AI should be leveraged to enhance decision-making processes, not just to generate more data.
“Giving human beings better insights doesn't typically lead to better decisions,” Hulm emphasises.
Lesson: Start by identifying decision-making problems within your organisation. Implement AI to improve these decisions, using algorithms designed to handle complex scenarios where human intuition often fails.
2. Recognise human limitations
Humans are inherently poor at making decisions involving many variables. AI excels in these areas, handling complex optimisation problems far more effectively than humans. Hulm notes: “I would argue you need to solve the problem first and work backwards. To expect a human to solve [exponential problems] we are wasting our time. Anything more than seven, don't use a human for.”
But he cautions: “If you use the wrong algorithm it will literally take longer than the age of the universe. The right algorithm will differentiate your business.”
Lesson: Deploy AI for tasks involving complex decision-making with numerous variables. This approach can achieve optimal solutions quickly and efficiently, freeing up human resources for more strategic activities.
3. Differentiate between automation and AI
Automation involves executing the same task repeatedly based on predefined rules. So we can get computers to do things like humans, for example to recognise objects and images or correspond in natural language like ChatGPT, but this is not intelligence. True AI involves goal-directed adaptive behaviour. These systems learn from their outcomes and improve over time, making them more suitable for dynamic environments.
Hulm argues: “Automation is amazing because we can get computers to do things better than human beings. But I would argue that by definition automation is stupid. It's not AI, it's not intelligence. Instead of using human beings as a pinnacle of intelligence. think goal- directed adaptive behaviour. Goal-directed in the sense you're trying to allocate your workforce to maximise wellbeing or route your vehicles to maximise deliveries or spend your marketing money to maximise return. Behaviour is how quickly and well you can move towards that goal.”
Lesson: Develop AI systems that are adaptive and capable of learning from their decisions. This adaptability will enable your organisation to respond to changing conditions and improve over time.
4. Focus on six key AI applications
Hulm identifies six categories of AI applications that can address any frictions in organisations:
1. Task automation: Replacing repetitive structured tasks to drive value
2. Content creation: Not creating generic content but generating brand-specific high-quality differentiated content
3. Human representation: Creating AI systems that mimic human perception
4. Insight extraction: Using machine learning to surface actionable insights
5. Decision-making: Applying optimisation algorithms to enhance decision-making
6. Human augmentation: Using AI to augment human capabilities through exoskeletons, cybernetics and digital twins.
“AI is not just good at creating content; we can now build brains and recreate how people perceive content. If I showed you an advertisement, historically I didn't know how you thought and felt about it unless I asked you – and people are not very good at reporting on what goes on in their minds and bodies. For the first time ever we can build synthetic audiences and get insights that we can now use to create better content but also now predict how that content is going to land,” Hulm explains.
Lesson: Map your organisational challenges to these six categories to identify where AI can be most effectively implemented.
5. Address AI risks proactively
AI implementation comes with risks that must be managed. These include ensuring that the intent behind AI deployment is appropriate, making algorithms explainable and preparing for both the positive and negative impacts of highly successful AI applications.
Hulm advises: “The difference between AIs and human beings is that AIs don't have intent. Human beings have intent. It is intent that needs to get scrutinised from an ethics perspective. If your algorithms are having a material impact on people's lives you can understand how they're making their decisions. Building explainable algorithms is extremely hard but necessary.”
Lesson: Establish a robust framework for AI governance that addresses intent, explainability and the potential consequences of AI outcomes. This proactive approach will help mitigate risks and ensure ethical AI deployment.
6. Empower talent and foster innovation
Successful AI integration requires accountable leadership, a commitment to innovation and a culture that empowers employees to utilise AI. Leadership must be convinced of AI's transformational potential and talent should be nurtured and trained to leverage AI technologies effectively.
Lesson: Invest in leadership development and employee training to build a culture of innovation. Encourage experimentation with AI and provide resources for continuous learning and development.
7. Leverage digital twins
Digital twins – virtual replicas of physical entities – are a powerful tool for optimising various aspects of an organisation. They can simulate supply chains, workforce dynamics and back-office processes, providing valuable insights and facilitating more informed decision-making.
As Hulm revealed: “One of the things that we're doing with one of the biggest brands in the world – this is going to sound really creepy – is that for each one of their team we create a large language model that's trained on their data, their email, their calendar, their communications, their feedback, all of their digital footprint. And we can ask that digital you, if I put you on this project, will you work well? If I put you on this team, will you thrive? And that's being embraced by those people because they feel like it represents them better than five numbers in an HR database.”
Lesson: Develop digital twins for key areas of your organisation to enhance predictive capabilities and optimise operations.
8. Navigate the future of work
AI's impact on jobs is a significant concern. While AI can displace certain roles, it also creates new opportunities for more meaningful and impactful work. Preparing for this shift involves reskilling employees and fostering a mindset that embraces continuous learning and adaptation.
“Yes, jobs will be displaced and be disrupted,” says Hulm, “but AI is like an energy source that will grow humanity – and it's going to provide a lot more opportunities for people.”
In the short term, if organisations quickly eliminate jobs to cut costs it could lead to mass unemployment and social unrest. However, there's another view that we should rapidly automate essential services like food, healthcare and education. By reducing the cost of these goods to near zero we could create a world where, despite the lack of paid work, all basic needs are abundantly met, he believes.
“If we apply AI in the right way over the next few decades we can free everybody from those economic constraints, allowing them to live their humanity.”
Lesson: Actively manage workforce transitions by investing in reskilling programmes and promoting a culture that values continuous learning. This approach will help your organisation adapt to AI-driven changes in the job market.
9. Implement agile and adaptive methodologies
Traditional hierarchical structures are not conducive to the rapid adaptation required in the AI era. Agile methodologies such as scrum and design thinking empower teams to innovate and adapt quickly, making your organisation more responsive to changes.
Hulm asserts: “The faster we can innovate, the more adaptive we are, the more intelligent we are. Agile methodologies empower teams and enable your organisation to adapt quickly to a rapidly changing world.”
Lesson: Transition to agile and adaptive organisational structures that promote innovation and rapid response to market changes. This flexibility will enhance your organisation's ability to leverage AI effectively.
10. Ensure AI alignment with organisational purpose
AI should be aligned with your organisation's purpose and values. A clear and compelling purpose will attract talent and customers, driving engagement and loyalty. AI can accelerate your journey towards achieving this purpose by enhancing efficiency and innovation.
As Hulm says: “I think a better world is a world where everybody's free to contribute to humanity however they want. It's you [enterprise] that will make this world a reality. It's that collective purpose of enterprise that will make the world amazing for all of us.”
Lesson: Define and communicate a strong organisational purpose that resonates with your stakeholders. Align AI initiatives with this purpose to ensure they contribute to your overall mission and values.
To thrive in an AI-driven future of work organisations need to meet three key requirements says Hulm The first is data. It’s data that makes AI smart. If organisations have data that contains insights their competitors don’t, that will differentiate them. The second differentiator is leadership. If an organisation’s leadership is not bought into the transformational power of AI, it is wasting its time. And finally, he says, the most important differentiator is talent. Empower talent to use these technologies to innovate and you will be a winner in the AI age.