麦肯锡公司将 2023 年称为生成式人工智能的突破之年，这是理所当然的。似乎没有哪个行业没有受到这项技术的影响。但是，尽管我们一直希望人工智能能够解决个人、社区和全球层面的挑战，但我们尚未充分利用这项技术来解决我们这个时代最紧迫、最重要的挑战之一：能源转型。 

国际能源署报告称，去年全球排放量创历史新高，虽然我们已经拥有许多必要的工具来支持安全、可靠和有保障的能源，但挑战在于实施。我们需要能源供应商的流程更加高效和自动化，而这正是生成式人工智能的用武之地。 

生成式人工智能的主要功能之一是分析大量复杂数据并将其转化为新的原创内容。其核心是，在正确的时间为正确的人提供正确的见解。

但要充分发挥其功能，组织首先需要释放通常锁定在不同系统和应用程序中的数据，从而使其基本无用。一旦数据更容易访问，就可以将其连接起来，以创建一个有意义且全面的视图，准确代表组织的工业现实。然后，这个工业知识图谱为大型语言模型提供必要的背景信息，并使数据可操作化，为公司提供做出更好决策所需的信息和背景信息。 

日本第三大炼油商科斯莫石油公司就是一个例子。几年前，该公司很难招聘到合格的工程师。该团队将此归因于日本出生率下降导致劳动年龄人口减少，以及石油和天然气行业在减排呼声日益高涨的情况下面临声誉挑战，导致招聘困难。 

当科斯莫石油公司的团队寻找一种以更少的工程师来运营炼油厂的方法时，他们最初的重点是弄清楚工程师如何开展他们的任务，以了解哪些地方可以最大限度地提高生产力，以及如何在多个站点之间整合运营。

研究发现，工程师大约 70%-80% 的工作涉及数据收集。这些关键数据包括运营数据、维护数据、设备检查记录、资产绩效管理工具数据、管道和仪表图以及各种组件的数据表，分散在多个来源和格式中。这些数据是孤立的，团队在试图追踪这些数据时浪费了时间和资源。

为了解决这个问题，科斯莫石油公司采用了Cognite Data Fusion，这是一个工业数据运营平台，可以收集所有以前无法使用的数据（包括非结构化数据），并使用人工智能整合和连接这些数据，实现数据自动语境化。从那里，该技术提取了灵活的知识图谱中的数据模式，以数字方式表示组织的运营。这使得科斯莫石油公司的工程师能够快速轻松地访问所有炼油厂的数据，从而简化数据驱动的工厂运营并提高整体效率。 

欧洲最大的独立石油和天然气运营商之一Aker BP和科技公司西门子都面临着类似的数据问题。Aker BP 已授权西门子访问其 Ivar Aasen 陆上团队的现场数据，并希望改善 Ivar Aasen 资产的状态监测。但西门子团队经常发现解决问题所需的数据被锁定在无法访问的系统中，阻碍了他们查看工作订单、工作许可和文件系统等重要信息。

西门子不再继续依靠零散且有限的洞察力进行运营，而是转向Cognite Data Fusion，该技术已在 Aker BP 的资产中投入使用。通过单一平台读取数据并通过单一入口点访问实时和历史数据（无论原始来源或格式如何）的能力节省了数百小时，并使 Aker BP 能够提高效率，降低设备故障成本并在需要维护时部署合适的人员和工具。

油田服务公司SLB看到了利用人工智能和数据解放来解决跨行业全球问题的机会，例如运营成本上升和减少碳排放的需要。数字技术是解决这些问题的核心，而数据是这些技术运作的核心。为了帮助上游参与者管理数据复杂性并就在哪里应用优化做出明智的决策，SLB 与 Cognite 合作，将公司的地下企业数据解决方案与 Cognite Data Fusion 集成在一起。 

通过当前的合作关系，SLB 地下企业数据解决方案的用户可以将来自油藏、油井和设施的数据集成到单一平台中，使其可访问，并使用嵌入式人工智能来识别优化机会和大规模创新的方法。 

该系统可以帮助公司超越解决现有问题，并为解决新兴用例奠定基础。 

从所有这些例子中，我们看到，工业数据管理与生成式人工智能的预测能力相结合，可以在简化密集数据流程、提高效率和可扩展性方面发挥根本作用。当我们寻求创造一个安全、可靠和可持续的能源未来时，能源供应商充分利用技术来创建高效和自动化流程的能力将至关重要。如果说 2023 年是生成式人工智能的突破之年，那么 2024 年就是它发挥作用的一年。 

McKinsey & Co. dubbed 2023 generative AI’s breakout year, and rightfully so. There was seemingly no industry that wasn’t touched by this technology in some way. But as much as we’ve looked to AI to solve challenges at the personal, community and global levels, we’ve not yet fully tapped into the technology to solve one of the most pressing and important challenges of our time: the energy transition. 

The International Energy Agency reported that global emissions hit an all-time high last year, and while we already have many of the necessary tools available to support a safe, secure and reliable energy source, the challenge lies in the implementation. We need energy providers’ processes to be more efficient and automated, which is where generative AI fits in. 

One of generative AI’s primary functions is analyzing vast amounts of complex data and turning it into new and original content. At its core, it’s a way to provide the right insights to the right people at the right time.

But to deliver on its full capabilities, organizations need to first liberate data that’s typically locked in different systems and applications, leaving it largely useless. Once the data is more easily accessible, it can be connected to create a meaningful and holistic view that accurately represents an organization’s industrial reality. This industrial knowledge graph is what then provides the necessary context to the large language model and operationalizes the data, arming the company with the information and context it needs to make better decisions. 

One example of this is Cosmo Oil, the third-largest oil refiner in Japan. A few years ago, the company was struggling to hire qualified engineers. The team attributed this to the decline in the working-age population as the birth rate decreased in Japan, as well as the reputational challenge oil and gas was facing amid growing calls for emissions reduction that made it difficult to recruit. 

As the team at Cosmo Oil searched for a way to operate a refinery with fewer engineers, its initial focus was on figuring out how engineers conducted their tasks to see where productivity could be maximized and how operations could be consolidated across multiple sites.

The research found that approximately 70%-80% of an engineer’s job involved data collection. This crucial data—which included operational data, maintenance data, equipment inspection records, asset performance management tool data, piping and instrumentation diagrams and data sheets for various components—was scattered across multiple sources and formats. The data was siloed, and the team was wasting time and resources trying to track it down.

To address this, Cosmo Oil turned to Cognite Data Fusion, an Industrial DataOps platform that could take all the previously unusable data, including unstructured data, and consolidate and connect it using AI for automated data contextualization. From there, the technology extracted data patterns in a flexible knowledge graph that represented the organization’s operations digitally. This allowed Cosmo Oil’s engineers to access data on all refineries quickly and easily, simplifying data-driven plant operations and increasing their overall efficiencies. 

Aker BP, one of Europe’s largest independent oil and gas operators, and Siemens, a technology company, faced a similar data problem. Aker BP had granted Siemens access to the field data of its Ivar Aasen onshore team and was looking to improve condition monitoring of the Ivar Aasen asset. But the Siemens team often found the data it needed to address the problem was locked in inaccessible systems, hindering their visibility into important information like work orders, work permits and document systems.

Rather than continuing to operate with fragmented and limited insight, Siemens turned to Cognite Data Fusion, which was already in operation across Aker BP’s assets. The ability to read data through a single platform and access live and historical data, regardless of original source or format, via a single point of entry, saved hundreds of hours and enabled Aker BP to increase efficiency, bring down the cost of equipment failure and deploy the right crew and tools when maintenance was needed.

SLB, an oilfield services company, saw an opportunity in the use of AI and the liberation of data to solve cross-industry global problems, like the rising cost of operations and the need to reduce carbon emissions. Digital technologies are central to solving these issues and data is central to the operation of these technologies. To help upstream players manage data complexity and make smart decisions around where to apply optimization, SLB partnered with Cognite to integrate the company’s Enterprise Data Solution for subsurface with Cognite Data Fusion. 

With the current partnership, users of SLB’s subsurface Enterprise Data Solution can integrate data from reservoirs, wells and facilities into a single platform, make it accessible and use embedded AI to identify opportunities for optimization and ways to innovate at scale. 

The system can help companies look beyond solving existing problems and build a foundation to solve emerging use cases. 

Across all of these examples, what we see is that industrial data management, in combination with the predictive capabilities of generative AI, can play a fundamental role in streamlining intense data processes to be more effective and scalable. As we look to create a safe, secure and sustainable energy future, energy providers’ abilities to fully harness technology to create efficient and automated processes will be paramount. If 2023 was generative AI’s breakout year, 2024 is the year of putting it to work.