How To Use ChatGPT At Work Without Getting Fired

Why Do I Need To Be Careful Using ChatGPT at Work?

ChatGPT is like that friend who's a sponge for information. It learns from the tidbits you toss its way. Super cool, right? But there's a twist: Imagine telling it a secret, and then it might (oops!) share that secret when chatting with someone else. A bit awkward for our confidential files, yeah?

Risky Business?

Many companies have stringent guidelines regarding the dissemination of their sensitive and proprietary data.

1. We all know how companies guard their treasures (read: data). It’s like their mom's secret pasta sauce recipe.

2. No company wants their secret customer list or game-changing strategy accidentally spilled.

3. Legal side note: Sharing some of this info with third-party platforms (yeah, like ChatGPT) might get some stern looks from the compliance and legal team.

Some Casual Hacks for Safe Chatting

😄 Just Explain the Problem:

Like telling a friend about a tricky Excel sheet without showing it.

Picture this: you're an analyst at a bustling property management firm. Your desk is filled with data, and your primary tool? Microsoft Excel. But there's a catch — your Excel proficiency isn't quite up to par with the complex tasks at hand.

Thankfully, there's a silver lining. With platforms like ChatGPT, you can get guided assistance. All you need to do is describe the Excel data you're working with, and voila! Let's dive into an example:

Prompt:

I have three columns, one for property values, rental prices, and property owner. I’m trying to aggregate potential rental revenue and property metrics: such as cash on cash return. What is the best way to approach this problem and what excel formulas would you write?

👨‍💻 Use Fake Data:

If you're working on something and wanna get ChatGPT’s take, just switch out real names and details for made-up ones. It’s like giving ChatGPT a mystery novel.

Got a proposal you're itching to run by ChatGPT? Cool, but first, let's talk privacy. You might wanna swap out those real names and company details for some made-up ones. Think "Acme Corp" or "Jane Doe." It's like giving your proposal a fun, harmless disguise!

For the tech wizards playing with code: Instead of spilling all the beans, how about sharing your code in 'pseudocode'? It’s like telling a story of what your code does without giving away the plot twists.

Got some data in a CSV? You can play two ways: One, you could change the sensitive bits (we're looking at you, Personally Identifiable Information). Or, for the super-cautious, whip up an entirely fictional set of data. Think of it as creating an alter ego for your dataset!

Bottom line? Share the fun, but keep the secrets, secret. Stay safe, pals!

Example Mock Dataset (CSV):

Name , Yearly Salary , Role 

John Doe , 10000 ,  Clown

Sally Jane, 5000, Ninja

👍Stick to the Basics: 

So, you've heard about metadata? Those nifty little details about the main data? Well, if you're sure they're not top-secret, you can totally slide them over to ChatGPT to give it a clearer picture. Think of it like giving someone the chapter titles of a book, but not the whole plot.

Working on a document? Just toss ChatGPT the titles or field names. No need to share the juicy details. Or if you're diving into databases, you can share the schema layout. Coding something cool? Give ChatGPT the function names but keep the magic (the actual implementation) to yourself.

Check this out:

Need help with an RFP? Just list the headings and let ChatGPT fill in the blanks.

Prompt:
Given a Request for Proposal With The Following Field- Please Fill in the Document. Background: I am a contractor with 5 years experience in building constructions location in Saint Louis, Missouri.
 
Fields:
- Title
- Summary
- Proposal
- Scope of Work
- Requirements

A Word of Caution

Though these workarounds offer a degree of protection, inadvertent slips could still lead to data breaches. And, while these methods ensure data security, they might not yield results as accurate as using real data.

To tackle these challenges, organizations are mulling over introducing proprietary LLM models or robust compliance software to scrutinize prompts. More on this will be explored in upcoming articles.

Stay abreast of these developments and more by subscribing to AI For Business, offering invaluable insights into seamlessly integrating AI into corporate environments.