ChatGPT: From Game-Changer to Game Over

As technology continues to advance, natural language processing (NLP) models like ChatGPT have gained significant traction. These models, powered by artificial intelligence, offer numerous possibilities in various domains.

ChatGPT, an impressive AI bot system developed by OpenAI, has the capability to provide optimal responses and fulfill user queries or requests in a remarkable manner. However, it's crucial to understand both their potential and limitations to make the most of their capabilities.

In this article, we will delve into three great use cases for ChatGPT in the ClimateTech space and three that should be approached with caution.

But before, let’s clarify a few key concepts:

👍 ChatGPT is great for 👍

Generating engaging and high-quality content consistently is a challenge for many individuals and organizations. With ChatGPT, this task becomes more manageable. The model can be trained to mimic the style and tone of a particular writer or brand, producing blog posts, social media updates, product descriptions, and more. By leveraging ChatGPT, content creators can save time and effort while maintaining their unique voice and brand identity.

Moreover, ChatGPT excels at synthesizing large documents. Whenever you need to convert an article into a tweet, summarize given text, or drive insights from a paper or even a book, ChatGPT is there to help!

The energy industry is facing problems of an aging workforce, talent acquisition, and employee training. In light of ambitious energy transition targets, having the ability to quickly upskill a team member is crucial for any company building and maintaining industrial assets - power plants, electrolyzers, carbon capture facilities, and so on.

Chat GPT has the potential to advance education by providing personalized learning experiences as well as synthesizing knowledge existing in the company. Integrating the language model into onboarding and internal educational tools can greatly accelerate employees’ readiness to deliver. Such a learning path can be personalized and can help people to learn at their own pace.

ChatGPT can also write code!

The significance of ChatGPT's code-writing capability stems from several factors. Firstly, it offers a substantial time and effort-saving advantage. Developing code, particularly for intricate algorithms, can be a time-consuming and monotonous task. By harnessing ChatGPT's code generation ability, programmers can allocate their time and attention to other critical aspects of software development, such as testing and debugging. This streamlined approach facilitates faster development cycles and enhances overall software development efficiency.

Secondly, ChatGPT's code-writing prowess contributes to improved code quality. Humans are inherently prone to errors, and these mistakes can easily find their way into the codebase. However, leveraging ChatGPT's code generation capabilities can mitigate such risks. By relying on an AI-powered system, the likelihood of introducing human errors is reduced, resulting in more reliable and robust code.

In conclusion, ChatGPT's code-writing capability brings significant benefits to the development process. It saves valuable time, enables focus on crucial development tasks, and enhances code quality by reducing human errors. By leveraging this advanced technology, programmers can expedite software development and ensure the delivery of high-quality, error-free code.

And that’s terrific for the digitalization of the energy sector!

One of the most impressive applications of ChatGPT lies in revolutionizing customer support and service. ChatGPT can provide instant, personalized responses, addressing common queries, troubleshooting issues, and guiding customers through their concerns. Its ability to understand natural language makes it a valuable tool for automating routine customer interactions, freeing up human agents for more complex tasks. Doing so can provide immediate responses to customer inquiries, reducing wait times and improving overall satisfaction.

As utilities become more and more truly customer-centric organizations, this tech will offer a huge productivity gain.

👎 ChatGPT is not meant for 👎

Despite all of the great use cases and impressive technology capabilities, ChatGPT is still a black hole. While for example in a Google search, you see where it's a source of information, with ChatGPT you do not. That means one needs to see the limits of this technology for today’s operational decisions.

We all know calculations, engineering, or solving math equations require deeper, logical thinking and looking at the problem from various angles. Yet, ChatGPT is a foundational, large language model, based on statistics and probabilistic - not logic. Chat GPT is really bad at complex engineering calculations and physics. Which is completely ok. It was not meant for that.

It excels at basically one thing which is it looks at a large amount of data finds patterns in those and then uses those patterns to make predictions. In the near term, it won’t replace structural design calculations or complex mathematical optimization.

But some AI developers still think ChatGPT might be able to perform such tasks in the future. We already see research on neural network models that incorporate physics equations. But it could take some time to fully operationalize this, and we still believe it wouldn’t be as accurate as a real person. All in all, we should look at AI as a way to augment humans and not completely replace their work.

As you're aware, Language Models (LLMs) undergo training on billions of parameters, but datasets of such magnitude are lacking in industrial settings. To ensure that Generative AI delivers accurate responses within industrial environments, it is imperative to establish an efficient method of equipping these AI solutions with contextual information. When referring to the context, I specifically mean a comprehensive and reliable digital representation of the industrial reality.

Applying any AI solution in the industrial setting requires a proper data foundation. A contextualized data model with DataOps principles in place.

Today, a good implementation of ChatGPT can provide employees with a natural language interface to locate relevant data from within an organization’s information systems. It must be said, however, that enterprise search tools already exist though, and they have been moving in the direction of semantic search, which helps to provide better results.

During project development, financing, or warrant arbitrage, legal costs can soar badly. And while ChatGPT can be an excellent resource for information, it is important to recognize its limitations in providing legal or financial advice. Laws and regulations in these domains are complex and constantly evolving, requiring expert knowledge and interpretation. Relying solely on ChatGPT for legal or financial advice can lead to misinformation or incorrect guidance. It is essential to consult professionals specializing in these areas for accurate and reliable assistance. That’s especially key for energy projects spanning various geographies, following specific jurisdictions and requirements.

It’s easy to fall into a trap here since ChatGPT outputs are unquestionably eloquent, which one may interpret as “intelligent”. But in reality, the model is repeatedly guessing the next word in a response based on a very large number of mathematical operations, resulting in a list of words and probabilities for each one. There is no true understanding of the subject like with us humans.

⏭️ Summing it up

ChatGPT holds immense potential in transforming various industries and it definitely has a role to play in digitalizing the energy sector. Its ability to engage in natural language conversations opens up exciting possibilities for customer support, content generation, and personalized virtual assistants.

However, it is vital to recognize the limitations and exercise caution in areas such as legal and financial advice, engineering design, and operational decisions. ChatGPT appears to optimize for plausibility rather than accuracy, and this plausibility can cause user complacency when assessing accuracy.

By understanding the strengths and weaknesses of ChatGPT, we can harness its power responsibly and unlock its true potential in a variety of domains, obviously also outside ClimateTech.