Artificial intelligence has long been the stuff of science fiction, but in recent years, AI has gotten closer to reality than ever before. From self-driving cars that make on-the-spot safety decisions to automated warehouses that intelligently route products from one location to another, AI is being used in exciting and surprising ways. OpenAI, a research and development company, has recently introduced ChatGPT to the world. This AI chatbot provides answers to questions and can even hold conversations, write research papers and output complex computer code in seconds. Its implications have been called everything from exciting to frightening, but it will undoubtedly have an impact on the future of automation.
The Current State of Automation
Before delving into how AI and automation work hand-in-hand, it’s first important to understand the current state of automation. Many automation systems rely on digital commands only, but some automation uses a combination of physical and digital components to issue commands. In the latter case, there is usually some type of manual intervention required to trigger an action, but in the former, processes can run day and night once a digital command sequence has been enacted.
Automation systems rely on programming that works by measuring tolerances. This is how AI “knows” what to do. When faced with a decision, AI programming looks at the available options that have been pre-programmed, and it then weighs the impact of each option within a set of tolerances in order to make the decision that best suits its anticipated outcome. In the case of an automated braking system, the AI will decide when to apply the brake and by how much based on how close a vehicle is to a barrier. The anticipated outcome is the prevention of a collision, so the AI will work through its available options to find the one that is most likely to result in this outcome.
How ChatGPT May Benefit Automation
Technologies like ChatGPT may benefit automation systems by providing an on-the-fly solution for making changes. Current AI automation systems require software updates and upgrades in order to change the parameters used to operate within. With a technology like ChatGPT, a user could potentially change these operating parameters to expand automation as needed, allowing an automated system to act as a modular unit that can be changed to suit changing needs.
Much like Arnold Schwarzenegger’s titular Terminator character in the popular film franchise, automation could be programmed to change its directives and functionality depending on the situation. This opens the door for automation and AI systems that are much more flexible and robust compared to what is available today. An automated warehouse could instruct its AI to look for specific things when scanning one set of products from an assembly line and then change this parameter to include additional details when a second set of products comes off the assembly line.
Home automation could also see a boost from technologies like ChatGPT. If an elderly loved one comes to live with a family member, home automation systems could have their parameters changed to adapt temperature settings to levels that are more safe and comfortable. Current home automation systems can have their programming set to adapt to environments and learn routines, but they can’t react in real time to changing variables that require complex logic.
The Potential Negative Impact of AI on Automation
Of course, the implications of ChatGPT on automation are not all positive. The potential exists for challenges to arise when AI programs hit a knowledge ceiling. In the case of ChatGPT, the program derives its seemingly endless knowledge from the Internet. As a result, it needs to have its language model updated on a regular basis to remain current. In 2023, ChatGPT is still working with knowledge from 2021, meaning it can’t output current knowledge. This can mean that automation system needs will be unmet if AI can’t keep up with changes.
Additionally, AI technology has logical limits that prevent it from understanding what it is outputting. These limits can be overcome, but this requires constant maintenance to update the core programming. This limitation has the potential to hold AI technology from reaching its full potential, and it may be a long time before this issue can be resolved.