In the first article of this series, we discussed how AI influences our daily lives and, in the second article, we learned how to handle the risks inherent in all AI, risks further highlighted by the incredible interest in ChatGPT and related Generative AI products.
Now it’s time to turn the spotlight on you. How can you use this technology, in your day-to-day work, to get the kinds of benefits we’re all used to as consumers, while taking the special care we must in any business setting?
Because none of us want to be those employees who sent proprietary company information to ChatGPT to help them summarize meeting notes, and check software code for errors … if they knew what you now know, maybe they would have thought twice!
Where to Begin
My first tip is to check your current company policy, which is likely to evolve rapidly over 2023. What data are you allowed to share externally, and what safeguards are in place? What does the company already recommend as best practice around Generative AI products?
And my second tip is, assuming you’ve checked those policies, to start with the parts of your job that rely on little or no proprietary information. For example, some salespeople at Dataiku are already using ChatGPT to help generate a crisp, one-paragraph summary of the hot issues in a given industry. Doing this shares nothing proprietary, because the fact that we target certain industries is no secret, and no customer or prospect data is shared in the process. If you think about any tasks within your job which boil down to “do a quick bit of research on the internet about some general topic,” perhaps you can save significant time!
Another favorite use case for the new Generative AI technologies is in graphic design. An events company we work with now generates most of their art using Midjourney, so the role of designers has shifted from hand-crafting absolutely everything, to generating many potential ideas with the assistance of AI, using their knowledge of the context to select the best matches, and then further customizing based on their knowledge and skills. This shifts their starting point drastically, and helps them get a new event advertised much sooner.
Moving Beyond the Chatbot
What these examples show is that any time we need to summarize a large amount of more or less generic information, and use that to generate something new, AI is already here. And we’ll slowly see the value of AI integrated into more and more other tools, be it Microsoft or Google, Salesforce or Hubspot, to help you take shortcuts. Just remember though: Shortcuts are never free … engage your brain every single time, the way we have become accustomed to scrutinizing emails or texts from the CEO saying they need to speak to us NOW.
Getting beyond using AI for shortcuts will take more effort, more time, and more care, because it will require either sending business data to third parties, or bringing new kinds of algorithms into your business. The good news is, any large enterprise has been doing this for some time, but perhaps just didn’t call it AI! This means that getting started with these more complex cases is easier than you might think: Much of the data already exists and the problems tend to be very well understood by the business — what has been missing is the ability to easily and safely apply time-tested AI techniques.
Practical Examples
A classic example is predictive maintenance in the airline industry. Current processes already generate a LOT of data, the challenge is that much of it is text notes from an engineer, with their own jargon. They have meaning when read by a single engineer standing in front of a single aircraft, but can generate additional value if aggregated to create a collective view of all engineers across all aircraft.
Giving individual engineers foresight of related issues that developed on similar aircraft allows them to schedule maintenance in advance, and ensure the right parts are ready on time, increasing aircraft availability and reducing delays and cancellations. The value-add of AI here is that it unlocks a new source of data for predictive maintenance, which is the combined wisdom of engineers in their notes, to augment the sensors already on the aircraft. And the AI technology to provide that augmentation existed long before ChatGPT made its splash.
Other classic applications of simple AI in airlines include predicting which meals will be preferred on any given flight, to reduce food waste and increase customer satisfaction, making the right co-marketing offers with credit card partners, to drive incremental revenue, and automating the routing of customer service messages to decrease time-to-resolution. The AI technology required to deliver this value is MUCH simpler than ChatGPT, but you need to apply it to your own data, because ChatGPT will not have a useful answer to these quickly-changing business problems. This is where platforms like Dataiku help reduce the entry barriers, and get those solutions out the door and generating business impact faster.
Wherever you start your AI journey at work, and no matter how quickly you move, I hope it is clear that learning about AI, the risks and opportunities, is crucial. And this is still a very human activity. I encourage you to take advantage of all the opportunities you have to get yourself ready to balance the opportunities (and risks) of this new AI age.