It's hard to ignore the level of excitement and interest around OpenAI’s Chat GPT, especially when you see some of the output it’s been producing in the ServiceNow space.
I've seen posts from my LinkedIn connections ranging from Business Analysts helping it to write test scripts, Product and Platform Owners using it to assist writing business cases and Technical Consultants and Developers to write or refactor JavaScript code. Just to test it out on something non-technical I even asked it to write a children’s story about a rabbit, it didn’t just produce a character that I could vividly imagine, but also a bunch of friends they were going to make mischief with and a fairly compelling story over several pages to boot.
Will robots take over our jobs?
As with the sudden awareness in most major advancements in technology, particularly when related to Artificial Intelligence, a lot of the talk then focuses on “if it can do all this where does it leave room for me and what skills I offer?”. For me the first thing to remember is that the source of information it has used to formulate its answer still comes from a fallible source, humans – though I recognise at some point it is going to be taking data from information it has created but we’ll have to cross that bridge later (particularly as I can’t get SkyNet, CyberDyne Systems and the Terminator out my head).
Be sure to do your own due diligence
If I focus on where I’ve seen the most comments and activity, which is the coding element, I still take the approach that as with any publicly available source you should still do your own due diligence before blindly using. When you get a code output from Chat GPT it really is then no different to looking at a community article containing script by someone who you see as a guru, but you should only be using that snippet if you yourself can dissect it and know what it is doing. I do admire it for giving you reasons as to why it’s given you the output it has so you can then rework what is provided to meet your needs. However, this is where you start to see the gaps, does it conform to your company coding conventions and standards? Does it prioritise readability over performance, or vice versa when required? Probably the most important thing is stating “you really should be using a Flow for this requirement rather than scripting”.
A template to work from
Its major benefit is giving you that template of information to work from, which is a huge help if you’re not quite sure where to begin if starting from scratch. One of our team found that if they could create a complex regular expression to analyse the data they were processing they could automate a very manual process. Not being fully versed in the dark arts of RegEx they were able to use Chat GPT to build it for them, refining it by keeping asking it questions until it gave them what they needed. That whole process took about 15 minutes and would have taken them 8 times longer without.
So there are huge improvements in productivity that can be made using this type of technology and it should be exploited, but you must agree the boundaries and terms of use in advance.
One final thing, would it surprise you if I told you Chat GPT wrote this entire blog post?
Just kidding…it was written by this entirely fallible human, but I bet I had you thinking!
