Don’t lock yourself into one solution for tackling a problem or workflow – design and build for change,
Menno Van Vechgel, IDT Manager, Information, Data and Analytics at Shell told the conference that flexibility was vital in such a fast-moving industry, given the possibility that current technology many already be outdated in a one or two years’ time.
“At Shell, we have learned from failures and successes, becoming more agnostic and building more flexibility into our Architecture at every layer. That means every component needs to become composable, whether it is an AI plugin, feature, workflow or it’s data. Using open standards is really key,” he said.
AI provides a set of tools that are changing the way the industry work and changing the pace at which the technology can provide value to operators, but there are dangers in deploying AI in the wrong context and through the wrong application.
Alessandro Speranza, Global Technology Portfolio Manager at KBC, told delegates:
“We shouldn’t be doing AI for the sake of doing AI. We should be doing AI when it matters, when it’s the right solution to our problem – starting from the problem you need to solve has to be the focus,” he said.
Leaders need to work on changing mindsets in their organisations. It’s not about whether AI or humans should be taking decisions, it’s about reinventing the way you engage with AI, Achilles Georgiu, Head of Data, AI & Innovation, MOL, told the conference.
“You need to empower your people not just to take advantage of AI, but also to be responsible for the outcome. This is not about AI, it’s more about the human side – how to reinvent ourselves,” he said.
Successful change management around AI-based solutions needs interaction with the ultimate users to get engagement and buy-in – and it has to be the right solution, Tyra Jones, Solution Engineer, EMEA, at Prometheus Group, said.
“You have to think about whether, once the solution is deployed, it is easy to use, easy to train, or easy to understand, Does the user need to go on a six- month course to learn how to use the AI? If so, then that’s not going to be very cost effective or productive for the business,” she told delegates.
The oil industry has been handling advanced process control and other automated technologies for years, and so is well accustomed to human oversight and responsibility on top of those processes.
That approach needs to be extended in the age of AI, one innovation and strategy manager at a leading operator told delegates.
“Using technology to get more automation will help us get to higher quality outcomes. But a human needs to be accountable for the outcome, if things go wrong. You can’t blame the technology,” he said.
People remain heavily involved in coordination of AI-driven processes – the humans-in-the-loop. But there is going to be a shift of emphasis according to Shane McArdle, CEO of industrial intelligence platform developer Falkor (formerly Kongsberg Digital).
“Human-in-the loop is the right model for now, but we need to move to human-on-the-loop, where people are managing self-improving systems – intervening only by exception, but remaining accountable for outcomes,” he said.
Projects will not move beyond proof of concept unless the underlying foundations are solid. For success, you need a clear data strategy, good governance, and good data ownership where your data is well managed in a repository that is actually accessible to AI, Shell’s Menno Van Vechgel, told the event.
“AI can run fast, but it is only as good as your data. Contextualised Data is the single most decisive factor for the success of your AI. You may see small successes, but most of them will stall if the data is not good enough,” he said.
Falkor’s Shane McArdle agreed.
“You don’t put intelligence on top of chaos. You need a solid data platform, where you build a knowledge graph of your organization and where you have access to all your data when needed – even if that data needs to remain inside their source systems to meet regulation,” Falkor CEO Shane McArdle said.
Digital Twins do a great job of organizing information, but AI still needs trusted operational context to make the right decisions.
Meaghan Snethlage, VP of Business Development, at VEERUM told the audience:
We need to adopt AI cautiously and progressively, but the question is no longer whether it will transform our operations—it already is. The real question is how well AI can understand our facility operations so it can deliver greater value and act with increasing autonomy in the future. Spatial and operational context are critical.
AI should not do your process hazard analysis for you, but it should make your job easier and more effective, according to Mariante van der Merwe, Product Manager, at Wolters Kluwer Enablon.
“AI is not suddenly going to start creating the entire data flow for your risk assessments – that will take work and effort, and it’s going to have legal and security implications. But if you have a connected platform then adding AI functionality becomes much easier. Trying to layer AI onto something disconnected is far more complex, because the platform itself isn’t set to support it,” she told the event.
Some speakers reported less impact on workforce numbers from AI implementation than they had thought might happen, with AI-based solutions replacing some tasks but creating a need for others.
“I’m becoming less and less concerned about jobs just disappearing – certain tasks will go or change, but higher productivity should also create room to employ more people in different areas,” Falkor’s Shane McArdle said.
Experienced hands can be cautious over the introduction of digital solutions, while young AI-savvy entrants to a company can be fearless about the consequences. Both approaches have their merits and downsides, so a mix of youth and experience on an operational team is ideal, MOL’s Achilles Georgiu told the conference,
“I think you need a good combination of skill sets. Experience is good but by itself can be a blocker, so it’s good to have that balanced by people who aren’t afraid of trying new solutions,” he said.
Collaboration within the oil and gas industry on data sharing and digital solutions has been easy to talk about but has proved harder to achieve. However, a number of organisations are seeking to change that.
Åshild Hanne Larsen, who is on secondment from Equinor as CEO of industry group CoOper8, said the benefits of digital collaboration were well documented, and would especially benefit areas where there are intense speed and cost pressures, such as maturing North Sea and Norwegian offshore fields.
“There is a lot of pressure on pace of development and cost that could be helped by collaborating more across companies. That is challenging, because there aren’t necessarily structures for shared incentives, but that’s exactly why the company that I’m now leading has been established,” she told the event.
CoOper8 has been set up by eight companies operating on the Norwegian continental shelf and in the adjacent UK North Sea to pool resources in order to improve data interoperability, deploy industry-wide solutions and cut costs.
