Recruiters, do not trust ChatGPT.
I know AI will impact us all.
And recruiting is no exception.
But you need to see recent AI developments for what they are;
The unfinished product.
Still in the test phase.
Prone to making massive errors
And also spreading total BS
I have two great examples to show you. Don’t miss these.
Let me show you an example of the ‘massive error’ first.
And it’s a key recruitment issue.
I asked ChatGPT to explain how to calculate a Temp margin for a Temp worker in the staffing industry.
(Please excuse all the typos in my question below, as I was just mucking around and not intending to screenshot this for a blog!)
However, it is the ChatGPT reply that I wish you to focus on.
The ChatGPT answer is just plain, deadest, 100% wrong.
That answer is incorrect. This is a concern because many (most) recruiters do not know how to calculate margin and get confused with the difference between ‘margin’ and ‘markup’.
ChatGPT made the rookie error of calculating the margin as a percentage of the Temp Pay Rate.
It is not.
It’s that percentage that the gross profit is to the Bill Rate. I.e., what the client pays, not what the temp gets paid.
To clarify. Bill Rate less Pay Rate divided by Bill Rate, e.g., BR =$30. PR = 20. 30-20 = 10. 10/30 = 33.33. The margin percentage is 33.3%. That is the gross margin because normally, taxes, superannuation and insurance will come out of that margin creating a lower % net margin.( Yes, about this, I know more than ChatGPT, and you can too if you subscribe to the Savage Recruitment Academy)
So, feeling weird, seeing as my modest human intelligence would contradict the almighty ‘Artificial Intelligence’, I told ChatGPT that it was wrong, and it very politely agreed and fixed the error. Or should I say, I fixed it?
But what if it was a legitimate query and I really needed to know?
Say, like many, I did not know the answer to my question. What if I constructed my pricing or quoted a client based on the word of the great AI oracle? What if I was a recruitment business owner unsure of the answer, who got help from ChatGPT and then trained the team on this wrong answer? It would be a disaster!
So that is the error I spoke of. Bad enough
But what about the BS, as I claim in my intro?
So, I decided to be a little frivolous with ChatGPT.
I asked it to tell me “Humorous remarks made by Greg Savage”.
I was fully expecting it to come back with nothing, because, you know, I am not funny.
But good old ChatGPT had plenty to say about my funniness. Apparently, I am a hoot!
Have a read below from the screenshot, and then make sure to read my response below that.
I have never in my life uttered these ‘jokes’. I have never said anything remotely like any of these ‘jokes’ in private, from a stage or in my writing. I have never heard these jokes.
I defy anyone to find any evidence of me having uttered these jokes anywhere.
But ChatGPT tells you that I said these things.
It’s totally fabricated. Utter BS.
It just grabbed some jokes from somewhere and attributed them to me.
And it is me, not some other ‘Greg Savage‘ with a dodgy sense of humour, because it mentions my recruitment background.
Now this does no harm. It’s just silly, and who cares if it quotes me falsely on this issue?
But in some situations, this could be disastrous.
Slanderous. Misleading. Lead to bad decisions.
It could mean someone does not get the job they deserve. Or a lot worse!
(Based on that report, I would not get a job as a comedian.)
The day before this exchange with ChatGPT, I asked it a legitimate question about how to treat a calf strain. It gave a convincing-sounding answer about ‘icing it’, and I acted on that. But is that BS too? How would I know? How do you know?
So, there are huge lessons here for all of us—particularly the AI acolytes.
Of course, AI is going to be huge. I am not contesting that. And it will improve fast.
But as a user, tread with caution. Don’t assume what it tells you is correct.
And as a consumer of other people’s content, remember they are increasingly likely to have created it via AI (But not this blog, baby!)
Which could lead to atrocious mistakes and misinformation.
For now, for sure, don’t trust ChatGPT.
*********************************************************************************************************************************
This is on now!
Nearly 1,000 recruiters attending
*************************************************************
AND NOW IN THE UK TOO!!
Registrations are now being taken. Six UK cities
*******************************************************************
- Posted by Greg Savage
- On July 17, 2023
- 9 Comments
9 Comments