Redesign Work, Rethink Productivity with GenAI

Food for Thought: Redesign Work, Rethink Productivity with GenAI

Senior technology leaders across industries gathered to discuss the challenges and opportunities of integrating generative AI (genAI) into their businesses. From workforce productivity to data governance, explore key insights and strategies from this Food for Thought discussion.

Businesses the world over have been experimenting with large language models (LLMs), the programmes behind the genAI corporate bloom, for well over a year. Whether to catalyse richer automation solutions or support young talent, aid in creativity or consider new revenue streams, the scope of potential is limited only by our imagination, we are told. 

In reality that is not the experience of executives. The technology C-suite is grappling with legacy architecture and opaque data systems from a technical perspective, and the abstract challenges of messy team culture and the idiosyncrasies of experimentation in a cost-conscious economy. This is before more universal factors such as regulations, ethics and governance are considered.

Still, that is not dissuading leaders from charting multiple paths forward with genAI—nor should it. 

The ability to redesign entire work streams and teams is too enticing a proposal; the emerging examples of genAI-enabled productivity boosts are a tantalising step forward in this “artificial revolution”; the positive experiences of the workforce already personally enjoying these platforms has a momentum businesses can propel. And at a time when the C-suite now see lack of innovation or creative ideas as a major concern within their team, redesigning work and rethinking productivity is an imperative.

Food for Thought, for the technology C-suite

It was because of this backdrop senior technology leaders from across the industry, from banking to pharmaceuticals, oil and gas, fashion and non-profit organisations, came together to share their experiences with genAI, compare, contrast, and learn from each other in a HotTopics Food for Thought lunch.

Across a two-hour, three-course lunch in central London, moderated by HotTopics’ Editor and Editorial and Strategy Director, Peter Stojanovic and Doug Drinkwater, leaders considered the AI integration already in their fields, the risks and controls both present and lacking, and the duality of regulatory frameworks alongside creative freedom and expression, amongst others.

Read further to understand these parameters in more detail, and to discover what your peers are thinking, doing and considering as they continue to experiment with genAI to ultimately improve their businesses.

This Food for Thought lunch was in partnership with Elsewhen. 

• Justin Gregory, Head of Technology, Payments, ASOS
• Mark Chapman, International Head of Investments, Nomura
• Peter Rees, Head of AI Data Platforms, Maersk
• Peter Krishnan, CDAO, JP Morgan
• Asier Nante, Global Head of Data, Sodexo
• Yashkaran Gill, Technology Audit Director, Morgan Stanley
• Simon Peachy, Director of IT, The London Clinic
• CK (Chuang) Kee, Director, Data Product, GSK
• David Reynolds, Director of Product & Strategy, AMS
• Roger Willmott, CTO, OAG
• Harinath K, Head of Architecture, Arthur Gallagher
• Leon Gauhman, CPO, Elsewhen

Food for Thought overview:

• GenAI and Workforce Productivity
• Control: A Defining Characteristic of GenAI Experimentation?
• Data Governance
• Big Pharma's Big Wins
• Role of the C-Suite

GenAI and Workforce Productivity 

Behavioural change management

The relationship between workforce productivity and genAI adoption weaved itself throughout the entire debate. 

For these leaders, a good working culture is necessary for any productivity gains to be realised. Yet, genAI does not provoke universal appeal: much of the work executives are expending today is around, for want of a better term, PR.

“Productivity is either efficiency or intelligence,” we heard. For either, genAI is both “an evolution and a revolution”, but convincing the entire team of this is proving difficult, as is reigning in those who are going off-piste in AI experimentation. A more cohesive approach to genAI as a tool is needed to maximise the experimenting of these technologies.  This can also be thought of as behavioural change management.

Some leaders are “struggling” here. Balancing traditionalists with techno-centrists was always tricky, but it has been made more complicated with the speed and success of genAI globally. But behavioural change management is also sector- and domain-specific, as we heard from the oil and gas industry.

The sector relies upon deep expertise in the life sciences, physics and mathematics, but challenges in the scope of new sites and in finding ever-increasing efficiencies for extraction mean “that will only get us so far today”. GenAI is seen as a “brilliance on top of all that”, a necessary augmentation to many aspects of the workforce’s responsibilities. That is quite the PR work for PhD-level scientists.

Augmentation was an agreed-upon term, however. “Experimenting with AI today is not to replace everybody with AI—we’re allowing a potential big picture of the future to stand in the way of building better solutions tomorrow. I’m pushing augmentation over replacement to get more buy-in.”

Technical skills gaps

“My biggest challenge is having the talent and time to build all our models that we need to demonstrate to the regulators and the vendors that these solutions are better for us and our customers.”

Therein lies the second major challenge from a workforce perspective. GenAI demands talent—specific, expensive and competitive. The backdrop of a wider talent shortfall and diminishing trust in the education sectors to shape tomorrow’s workforce does not help bolster confidence in the boardroom. It has created, as one executive commented, a “fragile” funnel of AI experimentation-testing-trialling-potential solution.

Does that encourage another look at the fixed structures of workplace hierarchies and ways of working? Potentially, yes. 

Creating successful genAI MVPs and use cases now involve bringing together technical experts, domain experts, and product and customer experts, many teams who do not historically work directly together. This is, amongst other things, exciting. The Covid pandemic saw unprecedented increase in the digital workload of employees that translated into productivity. The cost, as we are beginning to see, is a more jaded workforce more resistant to change and cynical of new enterprise tools that fail to live up to expectations.

Fragmented tool stacks, poor product design and a lack of integration into workflows have also compounded morale. New teams bring fresh ideas, which offer fresh perspectives. Upskilling and re-skilling is plugging gaps and fuelling a resurgence of pragmatic innovation. For some leaders, that is good enough for now. But for those in more regulated sectors, for example, the freedom to explore in this way is both potentially dangerous and counter-productive.

In short, for genAI to drive workforce productivity gains in its current guise, a business needs to consider its position in the market, and have an honest conversation about both its teams’ technical maturity and their psychological maturity. It is this combination of the objective and subjective that means no one path forward has emerged for leaders seeking a roadmap to success, yet.

Use Case 

Elsewhen recently collaborated with a FTSE 100 company to address their data analysis challenges. By leveraging generative AI and LLM technologies, they streamlined data processing in previously impossible ways, enabling new methods of analysis and data-driven decision-making. This ultimately elevated the client's business performance by providing its employees with an advanced capability.

Control: A Defining Characteristic of GenAI Experimentation?

As both an abstract parameter of experimentation and a required, measurable principle of security and governance, control was another hot topic. 

For the former, “control is mandatory for experimentation”. People are more creative when they have select rules in place, an aim or target, or lines in the sand. It explains when an empty white page can be daunting. The human mind counterintuitively finds some level of guidance more inspiring than none. In practice, fully understanding genAI’s capabilities allows for this inspiration. Leaders are seeing better AI traction when spending more time on education alongside experimentation for this reason.

For the latter, control is an industry or regulatory requirement. New solutions or modules are registered in an inventory, from ideation to production and beyond. Transparent inventories and the results of experimentation allows teams to check risks, and promotes trust within and outside of the organisation. (Trust is an under-appreciated quality of an innovative organisation, we heard.) This is working so far; keeping control is what is keeping leaders up at night. The appetite for experimentation is increasing, leaders reported, and the risk of shadow AI is a cause for concern.

To mitigate this, some executives have begun formalising genAI testing within their teams.

“We have had a big transition over the last 12 months to bring [different individuals and teams experimenting with genAI] into a more organised structure. We're looking, for instance, at a database rather than everyone popping up with their own capabilities, for example.”

Encouraging—or forcing—managers to apply good product design practice is also a clear option. What problem are you trying to solve? Is genAI the right tool to solve this problem? Who are you working with to make these judgement calls? Standardising a framework for experimentation was an agreed-upon next step for mature businesses in 2024. 

Data Governance

AI Acts had their moment in the debate.

Leaders, particularly those from financial institutions, complained of the confusion felt when dealing with legacy applications and data for their traders and investment bankers, and how, from an audit perspective, it all relates to upcoming laws. Not only are the security concerns pressing, but serious questions remain around if and how breaches will be penalised. 

“Are you comfortable that your AI framework will align with the AI Act in terms of security?” Few could answer concretely, especially when the possibility of Government change in the UK is being watched closely by industry hawks and veterans.

If banks and financial institutions are incredibly bearish on AI experimentation—one leader discussed the feasibility of their “two year plan for genAI”, to raised eyebrows—the pharmaceutical industry has been much quicker to gain ground. 

Big Pharma’s Big Wins

Pharma and medicinal research companies have had to comply strictly with research and patient data protocols for decades. Plus, the practice of drug discovery, for example, has not changed much at all in that time, meaning it has the rare breed of analytics: highly ordered data in large volumes. That has promulgated a bloom in big pharma-AI partnerships, ones that support intense automation of years of research, generative technologies that can seek novel pathways, and time-saving initiatives in an industry where making one drug can cost up to US$1B. In other words, exactly what other sectors are looking to achieve.

One of the leaders unpicked this somewhat, advising data federation as a model for others to investigate and replicate.

Data federation is essentially a software process that allows many databases to work together as one. Using this technology is useful for accessing sensitive biomedical health data, as the data remains within appropriate jurisdictional boundaries, while its metadata is centralised and searchable. Data federation is an “alternative to a model in which data is moved or duplicated then centrally housed—when data is moved it becomes vulnerable to interception and movement of large datasets is often very costly for researchers.” Instead, approved users may access the data via APIs. 

“It provides control, even in an environment where most organisations just do not know where their data is.”

“I’ve been [in the industry] for 35 years: we’re never going to get the data sorted,” came the response. New services, products, people, daily interactions—data is being created over a logarithmic curve and one leader in particular feels it is wasted energy to imagine all data being in an ideal format in the future. It encourages us, he added, to prioritise adaptability, resilience; qualities that can, say, react far quicker to sudden changes in AI Acts.

Role of the C-Suite

Leaders were split as to genAI’s revolutionary impact. It is a dichotomy we see played out in news, in the office and at home, too. All agreed, however, that leaders have a huge and unique role to play in curating effective, inclusive solutions and policies to benefit from and control genAI. In other words, recognition of bias in data is a priority for executives. In many ways the most important quality for C-suite executives today therefore is in storytelling.

Storytellers in this context map out the vision for the company and connect that to the drivers of business today, now including genAI. They intuitively sift through what is valuable and what is not, who is needed and who should be redirected, to fulfil that vision. They bring people along with them; they are empathetic to the needs of the workforce and they display a modern leadership style in the form of authentic, vulnerable and collaborative characteristics. In these nascent days of genAI leaders have more questions than answers. This is a novel situation but one that can and should be embraced, if only to support the spirit of experimentation and its better-known cousin, innovation, with something not often associated with the C-suite: humility.