AI Enthusiasts Also Need To Explore The Value Of Machine Learning

September 18, 2023  |  Mark Hillary

2023 has been filled with news about artificial intelligence (AI). The reason for all this attention has been the release of the OpenAI ChatGPT generative AI system in November 2022. It caught the public attention and was widely covered by mainstream media, rather than just the technology media.

It took just two months for ChatGPT to reach 100 million active users. That’s almost certainly the fastest adoption of any new app – even beyond popular games and social media platforms.

But tools like ChatGPT are trained and then released into the world. They are often upgraded, but the current version 4 is fixed in place. It doesn’t improve itself based on experience through a process of machine learning (ML). Users can’t improve it.

OpenAI, and companies with similar products, have suggested that this is because they don’t want a widely used AI tool to be shaped by the general public. This is because some users may deliberately manipulate the system to learn negative behavior. Google Vision Cloud has already been criticized for treating black and white people differently and openly available tools – such as Tay from Microsoft – have in the past been trained by some users to be racist inside 24 hours.

But the ability for AI systems to use ML to upgrade their knowledge through the experience they have can be important where it is used in a more controlled environment. There are multiple industrial use cases where a ML solution can create genuine savings and new opportunities, such as:

  • Manufacturing production line failure: use cameras and sensors to predict when line failures may be about to take place and to detect faulty products before they leave a factory.
  • Power line maintenance: using fixed cameras and drone footage the system can identify where power lines may be at risk of failure.
  • Object recognition from photos: a car brand may want to automatically detect how many of their vehicles are in use in a certain city or even which type of tires are being used – recognizing brands and specific objects can be automated with ML.
  • Customer service processes: AI-powered chatbots are getting better at having live conversations with customers and being able to fix their problems, however the addition of ML allows the customer service system to learn from every single customer interaction. In Europe, Atom Bank has been doing this for seven years now. Every customer problem and solution is learned by the system so the bot has the benefit of millions of customer interactions – only the most unusual of problems need to go to a human.
  • HR processes: many human resources processes involve data formatting, such as accepting CVs when job applicants apply and converting them into a standard internal format for comparison. ML can be used to teach the system these processes so many administrative tasks can be automated.

The press coverage of AI exploded in 2023, but the casual observer of how ChatGPT allows long question and answer sessions may not be aware of the importance of machine learning.

Although many of the public experiments in ML have been sabotaged by people who think it is fun to wreck artificial intelligence experiments, when used within a proprietary environment this can be a very powerful tool that allows your AI system to constantly learn and improve.

Imagine the value to your business of a system that knows about every problem every customer has ever faced or with such detailed knowledge of the production line that it can predict when there will be a problem.

Machine learning is an extremely important part of the AI story and most of the popular media coverage this year has ignored how it can transform AI solutions.

IBA Group has deployed many machine learning solutions. For more information on IBA Group’s ML expertise, please click here.


    Access full story Leave your corporate email to get a file.

      Subscribe A bank transforms the way they work and reach