No company can afford to ignore AI any longer. It offers plenty of opportunities, but there are still uncertainties too. Research conducted by Ipsos I&O on behalf of LINKIT shows that 57% of IT decision-makers expect AI to have a major impact on IT capacity.
In this article we cover:
- Training your people: from AI literacy to awareness of the risks
- Getting your data ready: quality, governance and architecture in order
- Implementing AI: building, integrating and scaling use cases
- Throughout: managing the risks and the value of external expertise at every stage
Training your people
The work employees do is changing. That applies to people within IT, but also well beyond it. AI is becoming part of everyday tasks. Niels de Bruin, Head of Commerce at LINKIT, sees that there is still a lot of uncertainty about which competencies are needed.
“Many organizations are struggling with what knowledge they should be giving their employees. We have to accept that, as a market, we are still learning.”
To bring employees along, that knowledge has to be transferred in an accessible way. So start with the basics. How do you write the right prompt? How do you handle data, and which data is suitable to use? By offering internal training, employees can grow along with the development of AI and learn to judge AI on its merits: not a threat or a miracle cure, but a useful tool. And it creates awareness of risks such as bias and privacy.
Skipping that basic step and hiring an AI specialist straight away? That is often not advisable. First take stock of whether your employees and your relevant data are ready to start working with AI, and what exactly you want to use AI for. An expert cannot get to work in an organization that is not ready to implement AI. Niels:
“Don’t expect to bring in a guru who will set everything up for you.”
Getting your data ready
In the preparatory phase, external expertise can certainly be useful: in change management, or to transfer knowledge to internal employees. Data engineers and data governance specialists can also help to get the relevant data ready, check the quality of that data, and draw up policy for using that data and AI. Many existing IT foundations fall short on scalability and speed, and revising data architectures is a job that is easily underestimated.
Security plays a part here as well: where does the data you enter end up? Particularly with AI systems that come from the United States, this is a question the public sector in particular is wrestling with.
Implementing AI
Only once all of this has been worked through can an organization build applications, integrate them into existing systems and finally scale them up. That calls for both process knowledge of the organization itself and technical AI/ML expertise.
One of the big changes for the tech sector is that AI can also write code, and can therefore build applications too. That sounds attractive: it is fast and requires less specialist skill from the maker during the build phase.
But handing over the coding work does create new risks when something has to be changed after delivery. The organization also loses sight of how an application has been put together. If the application grinds to a halt or needs to be extended, there is no one who knows how the code fits together. Niels:
“Especially when it comes to core applications, that is highly risky.”
On top of that, costs will rise sharply in the years ahead. Organizations are integrating AI into their processes now, but lose sight of what the future costs will be. Niels expects those costs to become many times higher.
“If organizations don’t factor that in, they really do have a problem.”
From time saved to real value
Niels:
“Companies that do nothing with AI will have no right to exist in the long run. But neither will companies that do it badly.”
AI is an important tool for automating processes and making them more efficient. That makes it possible to win back time and capacity, provided it is integrated in the right way.
One way is for a developer to use AI to generate code, but to review and implement that code themselves. The development process speeds up, while staying safe, manageable and affordable.
TNO carried out research into the added value of AI in business, with LINKIT contributing. An important conclusion from TNO is that time saved through AI only becomes a real gain if an organization fills that time meaningfully. It can mean employees experience less pressure, which benefits their mental wellbeing. A team can use the freed-up time for additional tasks or to work more cost-efficiently. And the gain can just as well come from an increase in quality.
The value of external expertise in practice
Nigel van der Wurf is an account manager at LINKIT, working with a large organization in the tech industry among others. That company is investing heavily in AI and the first programmes are live: automating the scheduling of job interviews, for example. AI applications that affect processes more fundamentally are still in development, Nigel says.
He sees a shift taking place from data engineering to AI engineering. The company is investing in training internal employees in how they work with AI, but not all knowledge can be found in-house.
The basis for a new concept always lies within the organization, but they do bring in external specialists to think critically about the ideal outcome. Those specialists stay involved while the product is developed and work alongside internal employees, so that the knowledge is anchored within the organization. Finally, the organization places the management of the application with off- and nearshore partners. That is how they use resources and knowledge efficiently.
In general, this organization brings in external expertise to build knowledge within the organization, or to scale up capacity temporarily. This is how the company sets up all its projects, but AI has made it more relevant still. AI expertise is expensive. By hiring deliberately in the different phases, so that everyone does the work they are best at, the organization can keep costs down.
But scarcity brings challenges of its own, Nigel adds.
“It is simply not always possible to hire someone internally with the right expertise. That is when hiring in is a valuable solution.”
Conclusion
Broadly speaking, an organization goes through three phases: first training the people, then getting the data ready, and only then implementing AI. AI is not a holy grail with limitless possibilities, but a tool that can deliver major benefits when it is used deliberately. That calls for a phased build-up of knowledge and technology. It matters that internal employees have a solid base of knowledge, both of the opportunities AI offers them and of the risks it brings.
AI is only as good as the data underneath it. It is up to an organization to make sure that data is usable and of good quality, but if the expertise for that is missing, hiring in external specialists can be a solution. To make those preparations and call on the right (external) expertise, a clear vision is essential.
Only once the foundation is in place can the building begin. An organization can work with external partners to integrate AI expertise into a team. That way an organization acquires the knowledge it needs to build and implement applications.
Would you like to learn more about how organizations manage IT capacity and where the biggest challenges lie?
Explore all the insights from the LINKIT and Ipsos I&O study, or schedule a no-obligation session with our experts.
