Business Actions Re: ChatGPT You Should Take Right Now

If you’re in a leadership position the rapid rise of ChatGPT and similar tools means there are actions you need to take right now. This is another Google or iPhone moment. If you’re still in “wait and see mode” I hope this article nudges you out of it.

Housekeeping Note: This was going to be my 3rd and final post on the impact of GPT and other Large Language Models on technology and business. As I’ve dug deeper, I’ve realized there are still too many topics I want to discuss. Treat this as 3a and please consider reading my previous articles What ChatGPT Means for Engineers Today and The Robots Aren’t Coming for the Testers .

While I intend to explore the future business impacts, I see from these new AI tools, I think there are important things that business and technology leaders need to do right now. These are important to protect your business and customers and to set you up for success. Do not wait.

1) Set explicit information governance rules and train everyone, immediately

This is first because I don’t think the average user of these new AI tools is thinking about security or privacy at all. ChatGPT is not a safe place to enter any of your important company or customer data.

You don’t have a BAA with OpenAI to pass in HIPAA data. You don’t have an NDA where you can rely on them to not leak PII (personally identifiable information) about your customers. You absolutely must not give GPT any non-public customer info.

It’s also not a smart place to pass in your key business strategies. The model may not use that as training data today, but I expect it will before long.

Schedule an all hands call and discuss these points

It is never appropriate to give private information external AI. Giving PHI (protected health information) or PII to an external AI will open your company to lawsuits and may be a terminable offense.
It is appropriate to use AI as an alternative to search engines, but it’s never appropriate to blindly trust the results for decision making
Current AIs make confident and reasonable sounding mistake. The AI equivalent of SEO to tamper with results is already being explored.
No code generated from AI should be run without a programmer reviewing it. We haven’t heard about GPT injection attacks yet but I bet we will within 3–6 months.

Talk about this in your 1:1s too. Things are moving fast and you should not expect all your employees to know what’s happening.

Nobody should get to skip the training, not even executives. In my experience, executives can be as guilty (or more so) as anyone else at falling to phishing scams. I think AI data leakage will be similar.

2) Train every employee and enable them to experiment

Now that the scary part is over, I suggest you also start training your employees on the basic uses of ChatGPT and building AI prompts soon. If you haven’t experimented yourself, I recommend you do so.

Ask your team members if any of them have used it and promise they won’t be in trouble. Anyone who has used it heavily for a week or two can likely do a basic intro training on how the ChatGPT works. Use that person as a seed of knowledge in your company and carefully go over point 1 with them.

ChatGPT has replaced Google for half of my searches now and it’s sped up almost everything I do at my computer. If you don’t have anyone on your team who can do this training, leave me a private note (see above) and I’ll be happy to help.

Not sure how to leave a private note? Highlighting any text in this article and click the speech bubble button you’ll see pop up. Please include your email address so I can follow up!

3) Reconsider and slow your hiring plans

I expect this to be contentious but here goes: if you are about to hire another person to expand an existing team in a creative or analytical role, pause. If you desperately need to hire someone, do it, but if a team of people is at or near their capacity to do work, they might not be for long.

The efficiency gains from using GPT and the LLMs that will be in the market soon are huge. Imagine hiring more typists and copy editors and then having someone give you Microsoft Word or Google Docs. Or imagine hiring a few librarians/archivists and then being handed Google and Wikipedia.

I don’t think AI will take all our jobs but I do think we’re going to see a major and messy shift of priorities. You don’t want to hire someone who becomes redundant in two weeks. Pause, think through the impact to your business if every employee working on a computer gets 25–50% more effective this year. If the answer is still “we need to hire more” then do it!

Wrapping Up

I want to get back to deeper thoughts about the impact that GPT and other LLMs will have on business but the more I thought about the risks of data leakage, the more I thought an immediate and tactical post was in order.

If you appreciate these articles, please like them and follow me on Medium. Thank you!