AI has changed the landscape of our professional and daily lives. Need a quick email written? You can turn to AI. Want a healthy meal plan? Ask AI. Stuck on a catchy title for your blog? Look to AI for suggestions. Most people, to varying degrees, have found ways to incorporate AI into their lives. But have we stopped to consider how we can use AI responsibly and the potential drawbacks and limitations of its use? In tis blog, we will provide an intro to AI safety so you can leverage it effectively.

No matter how helpful you have found AI to be, the fact remains that it is still a young technology. Some of the knowledge AI operates on is outdated because it has to take the time to learn it from existing sources. We aren’t at the point where AI can generate entirely new ideas – it still relies on people to come up with the ideas and then learn from them. 

We are seeing technology change rapidly and it will continue to do so. Asking AI to keep up with these rapid and complex changes is difficult, which is why it is notoriously behind on updates.  

Intro to AI Safety: A Real-Life Example 

I sat down with one of our developers, Steve, to talk about some AI challenges and he provided me with a couple of examples of where he has encountered issues with AI.  

He explained that in Power App development he would ask AI to solve an issue and to provide a line of code. What it gave him worked in Power Apps two years ago but doesn’t work that way anymore. If you were to use the code provided by AI, Power Apps would have thrown out an error. Microsoft has changed and evolved, and AI has yet to catch up. It pointed Steve in the right direction, but didn’t provide him with the correct answer.  

The second example Steve provided was that ChatGPT instructed him to use the “Union” function in the app he was working on in PowerApps.  The issue is that you cannot use “Union” in PowerApps as that function in only for Power Automate. The information was simply incorrect and not useable.  

Intro to AI Safety: True Information vs. Untrue Information

AI is still in the learning phase, and it likely will be for a long time. It is constantly learning. The machine learning algorithm is only as good as the information that is currently available. Not everything on the internet is true. Just because AI grabs it, doesn’t mean it’s automatically correct. A good rule of thumb is to take everything AI says with a grain of salt. Don’t blindly put your faith in it.  

You can think of AI is the next generation’s Wikipedia. You can use it for research purposes, but you always want to fact-check the information and you never want to cite it as a legitimate source.  

10 years ago, if you had used Wikipedia to defend a point or an argument, people would have scoffed and likely brushed your argument off. The same can be said if you were to use AI in the same way today. 

The point of this blog is not to deter you from using AI. There are many positives of using AI and it has changed the way in which we find and use information. You should use it as a tool, but you need to be aware of the setbacks and you should always tread lightly. Don’t take everything AI says as gospel because it might not be 100% accurate.  

Thank you for reading! If you are curious about how your company can leverage AI effectively and how to use it responsibly, get in touch by filling out the form at the bottom of this page.  

Make sure you subscribe to our blog. We publish technology tips, tricks, and updates every week.  

Discover More