Artificial Intelligence (AI) is touted as an enormous opportunity for all industry segments, from consumers to businesses to healthcare and beyond. A recent introduction of “ChatGPT,” a free artificial intelligence chatbot developed by OpenAI, released in November 2022, has captured the public imagination in a big way.
Artificial Intelligence (AI) is machine-based intelligence where a system or a program learns from intensive iterative training and feedback after each experience. We deal with AI-embedded computer programs every time we purchase a product from an online store when the webpage provides the shopper with additional suggestions based on the purchases by other customers or the shopper’s purchase history. Sometimes it can be convenient, but sometimes it can be annoying. SIRI and ALEXA are based on an AI platform. They ‘know’ your preferences and remember them. Have you noticed that your “Outlook” email system has been upgraded to include automatically finishing your sentences and your Word app? Finishing another’s sentence has gone beyond the long-time married husband-wife sentence ending.
Everyone in the C-suite has AI on their mind and lips in front of investors. But how can it be of use in your industry? Here are some basic suggestions:
- Predictive Maintenance
- Quality Assurance & Inspection
- Demand Forecasting
- New materials/product development
And the list goes on. Every day, new applications are proposed or introduced by AI companies. These can include advances in machine learning, neural networks, deep learning, big data analysis, machine/computer vision systems, robotics, face recognition, gesture recognition, and more.
Depending on the complexity of the problem, developing an AI methodology can require–
- Millions or billions of training examples
- Many training cycles
- Retraining when presented with new information
- Many weighted functions and lots of multiplication.
Everyone tells us that AI is so good. What could go wrong? Quite a bit. Poor quality input data or erroneous data can significantly skew the results and introduce statistical bias in the system. We take it for granted that when we punch in an address to our GPS unit, it will get us there. If a bridge is out, but the data was not entered into the AI algorithm, we could literally fall off a cliff.
AI companies make it sound like AI is a plug-and-play black box, but there is much groundwork internally you need to lay first.
- Is your management on board?
- How realistic are your expectations?
- Do you have the budget?
- Do you have the right people on your team to shepherd the project?
- Have you defined success metrics before jumping in with both feet?
- Are you willing to pull the plug if that black box is not working for you?