ChatGPT is making headlines. OpenAI recently released its latest version of ChatGPT, which can provide “human-level performance” on many professional tests, such as the bar exam and the SAT.
But what can it do in the industrial space? Here we chat with Mikaela Pisani, head of data science for Rootstrap, about the newest version of ChatGPT and how manufacturers can actually use the ongoing development of new AI technology to improve business outcomes. Take a look…
Smart Industry: How is ChatGPT changing manufacturing?
Mikaela: ChatGPT might change the manufacturing industry in many ways, creating optimizations plans, predictive-maintenance schedules, mitigating risks, improving communications (making them faster and more efficient), as well provide quality control-detecting anomalies in the available information.
Manufacturers using this tool would be able to make better decisions, reduce costs, improve quality and, as a consequence, have a greater customer satisfaction.
Smart Industry: How does it change the relationship with customers?
Mikaela: The chatbot can provide quick responses for common issues that customers have, enable faster diagnosis and suggest personalized recommendations. Therefore, it can optimize time in responses giving the user useful information helping manufacturers to build stronger relationships with their customers by providing better service, personalization and responsiveness.
Smart Industry: What concerns/challenges does this tool present?
Mikaela: While ChatGPT and other AI technologies offer significant benefits for the manufacturing industry, they also present several concerns and challenges. AI systems might be vulnerable to cyberattacks, compromising sensitive information. It is manufacturers´ responsibility to ensure security over AI systems.
Also, there is a trend that more and more AI is replacing automated tasks, and without a human in the loop there might be ethical concerns regarding decisions made with these tools. Therefore, manufacturers should consider introducing humans to validate and correct AI results in order to provide a secure service.
To ensure privacy, data capture should be performed, considering relevant regulations and ethical standards to be used as input for AI algorithms.
In addition, a big challenge for manufactures is the integration of AI systems into the common workflow and processes. The use of AI should be done by steps, enabling compatibility with other systems.
The workforce is not familiar with this technology, so owners need to invest money and time into training workers to integrate the tool in their job. On the other hand, there might be a reduction of personnel, since AI can automate many tasks and processes that are currently being done by people; it might be better for the company to hire trained people rather than wait for the current workforce to be ready for the job. This may lead to job displacement.
Smart Industry: What limitations exist with ChatGPT for application in the industrial space and how long will these limitations exist?
Mikaela: The limitations that ChatGPT have are the following:
1. The lack of actual information. This fact is a problem when you need updated information, and also this model is not very useful when you need to validate the source. In the short term this won´t be a problem, since Microsoft and Google are realizing search engines connected with a language model like this. Therefore, the mix of a search engine and a chatbot will be really powerful tool for searching information fast and getting a summarization.
2. It does not have any real-world experience, so the understanding of certain situations might be limited. This might be a problem because in some cases, practical knowledge and skills are essential for success in the field. This limitation will persist for a long time, because it is one of the elements that separates us from machines. It is true that machines can have certain memory and learn from the experience and simulation, but their experience is more limited than ours. It may not have the same level of intuition and creativity that a human would bring to a problem. So it is unlikely to fully replace the benefits that come from hands-on experience in the industrial space.
3. Bias: AI models present bias inferred from training data or patterns that they learned. Bias can affect the responses and decision-making. Bias will be always present since it is impossible to remove it; even when we think that bias is zero, there will be bias. This fact is not that fatal since we already have human bias; maybe with the help of AI we are able to reduce it.
Smart Industry: What does the use of this tool look like five years down the road?
Mikaela: I feel that this tool is revolutionizing our way of work. We will need it to make our work faster and keep up with market expectations. In the near future, AI might be used on every step of manufacturing process, in this way manufacturers will exponentially improve use of resources and time, focusing on providing better quality and freeing more time for working on creative ideas that deliver innovation.
In terms of our society, in five years I think that there will be more personalized tools that are branches from this one, and our lives will be more automated and optimized.
Smart Industry: What do you find encouraging about this tool for the manufacturing arena?
Mikaela: In my opinion, there are a huge amount of opportunity to take advantage of this tool in the manufacturing area—think automating communications, optimizing costs and resources, providing opportunities for workers to focus on more strategic tasks rather than repetitive tasks, to detect errors and predict maintenance, as well as provide useful insights for decision-making and sharper recommendations.
