[QTalks Ep.14]

The momentum surrounding Artificial Intelligence (AI) has never been higher, and the arrival of ChatGPT has all industries considering what the future of work really looks like. The question is, can ChatGPT make the process of water management easier? Is it just a productivity-enhancing tool? Or, will it change the water sector forever?

Discussing how is AI reshaping the future of water management and empowering utilities to make informed decisions, the risks involved in AI-enabled water management, and much more, our panel joined environmental journalist and QTalks host Tom Freyberg to answer these questions:

Full episode available below.

Are water-focused digital start-ups at risk of replacement by AI language models?

Tom kicked off the discussion by posing a statement to the panel — “Water-focused digital start-ups risk being put out of business unless they quickly learn how to integrate open AI with their solutions” — and asking them to elaborate on whether they agree or disagree with the sentiment.

Answering first, Gigi agreed with the statement. She said that every single business should understand what large language models can do and how they can leverage them and associated plug-ins. Referencing answer engine WolframAlpha, which large language models and other plug-ins can access, Gigi said that this could begin to replace some of the feature sets that small start-ups have been providing. 

Saša also agreed with the statement and extended the sentiment to any content generation role including text, graphics, and video. He said that ChatGPT and similar tools will fundamentally change how we create content by adding the value of productivity and that while ChatGPT won’t replace content generation entirely, those who harness it efficiently will have an advantage over those that don’t. 

Christos then said that he wouldn’t disagree with the statement and that the “efficiency multiplier” and productivity effect is what is most interesting about AI tools. He said that achieving a prototype with the aid of these tools can be done very quickly by using them to capture requirements from physical language and then turning it into code.

ChatGPT in use: Using the tool to emulate a senior water engineering consultant

Tom then introduced a guest panelist — ChatGPT itself. Tom entered a prompt into the ChatGPT engine, asking it to act as if it were a senior consultant with 20 years of experience in water engineering. He also asked how the “consultant” views the potential of ChatGPT4 and the integration of plug-ins to help improve the delivery of water services. 

ChatGPT’s “senior consultant” responded by saying that ChatGPT4 has the potential to revolutionize the way we deliver water services. The engine provided examples, including the potential to develop plug-ins that enables ChatGPT4 to analyze data from water quality sensors or weather forecasts, providing real-time insights into the condition of water systems and potential risks. 

Reflecting on this response, Saša said he believes that AI provides the ability to look at data in a different way. Whereas traditionally, we had the statistical analysis of data and deterministic modeling, Saša said that even though they’re a black box, large language models combine both of these approaches to identify new patterns that are missed by using mathematical and statistical models.

ChatGPT and water management case studies

Tom then raised the need for more use cases where AI is having a positive effect throughout the industry. Gigi said that she has already used Google Colab and introduced an environment to bring in OpenAI and enter a data file with 637,000 lines of AMI data that was processed within seconds. She was able to see specific household data and access multiple types of statistical information rapidly. 

She also said that this is a powerful tool in terms of the data democratization process. Imagining a scenario where utilities around the world inputted AMI data and metadata, she said that this would result in a large amount of extremely beneficial data that would help in creating virtual sensors. 

Christos said that the uniqueness of AI is that large language models are an interface between the vagueness but also the richness of physical language. He said that he predicts that we’ll soon see use cases for translating the data to real-world scenarios, in essence bridging the gap between the data and human understanding. He said, for example, we might see more interactive training scenarios for operators.

What are the potential risks associated with AI-enabled water management?

Tom rounded off the discussion by asking the panel about the downsides and risks of AI-enabled water management that utilities and companies should be aware of. 

Saša said that we must be careful of blindly accepting the ChatGPT results as true. He said, for example, that it’s important to be aware and sure of the assumptions and demands that are inputted into the engine, and double checking these against the output it creates. 

Gigi echoed this sentiment and said that the right guard rails must be put in place around large language models. She said that general AI is on its way to being “smarter” than humans and that we need to understand how we’re going to coexist with AI systems. 

Christos said that we must also be mindful of the issues around ownership and trust in terms of shared data across large language models. Secondly, he said that companies must also be aware that prompting a machine to act like an expert can only partially replace the in-house expertise they already have. He said that the rise in AI will also have an effect on how universities will educate young engineers. 

Finally, Christos said that the issue of misinformation across mission-critical sectors like the water industry is huge and that they are targets for cyber attacks. He said that we need to be aware of how powerful AI engines can also be used for nefarious attacks and that the industry needs to consider safety and cybersecurity as a fundamental priority.

Ready to discover more QTalks content?

If you’re looking for more insight on how water utilities can take action on AI and water management, check out Qatium’s advisory board white paper: AI & Water management: What utilities need to know now

Plus, visit Qatium’s YouTube channel to watch this episode and previous ones.

You might also like...