Brilliant video. I have had a lot of success with ChatGPT 3.5 but it took a lot of practice with prompts and I found what you did with table and columns names is the key. It is important to stay consistent with the naming so a model can be determined. It made me think though that it will only be a matter of time that the language model sits natively in BI so it knows the model already. Then they release Co-Pilot! Amazing.
Really love where you’re going with this! You are convincing me to buy the upgrade. The point is, yes, we have to know DAX, but what a timesaver it is to have the pre-written prompts to getting the code written in clean format and commented on, all at once. Then all we have to do is test it, and make the necessary tweaks. Love it! Thank you again!
Very nice video, and thank you for pointing out that ChatGPT isn't a substitute for learning DAX. People at my work often take shortcuts and think they can ask questions without having any context on the subject matter. I guess they are simply too lazy to learn. I use ChatGPT as a tool to help me, but I believe that learning the subject matter is important. For instance, I have to learn DAX and not rely 100% on ChatGPT.
Great video, thanks. As a trainer I sometimes have difficulty to explain difficult DAX formulas to trainees. I guess I could use GPT4 to get explanations of self-built formulas.
I think it’s initial impact will be to massively improve the efficiency of those that use it. It will also reduce but not remove the need for many specialist areas, which has been steadily the case for many years now.
I tried it with Google Bart and it came up with the following DAX: // Create a variable for the reservations table var reservations = CALCULATETABLE('Reservations'); // Create a variable for the calendar table var calendar = CALCULATETABLE('Calendar'); // Create a variable for the number of rooms booked on each day var roomsBooked = CALCULATE( SUM(reservations[Room]), // Filter the reservations to only include days that are in the calendar table WHERE( reservations[Arrival]
GPT is nowhere near to being as good at DAX as it is with other languages because DAX is typically contained in PBIX files and those are not source code. There are barely any GitHub repos containing DAX code. The model could never learn it properly.
So helpful, practical and with examples and a template how to use - brilliant
Cheers Justin
Brilliant video. I have had a lot of success with ChatGPT 3.5 but it took a lot of practice with prompts and I found what you did with table and columns names is the key. It is important to stay consistent with the naming so a model can be determined. It made me think though that it will only be a matter of time that the language model sits natively in BI so it knows the model already. Then they release Co-Pilot! Amazing.
Thanks. Yes it will be very interesting to see how co-pilot develops
Really love where you’re going with this! You are convincing me to buy the upgrade. The point is, yes, we have to know DAX, but what a timesaver it is to have the pre-written prompts to getting the code written in clean format and commented on, all at once. Then all we have to do is test it, and make the necessary tweaks. Love it! Thank you again!
It’s hard to imagine where this will take us in the next 12 months
Amazing 😮
It is quite impressive. Great video cheers.
Thank you
Excellent video Wyn!
Cheers Chris
Very nice video, and thank you for pointing out that ChatGPT isn't a substitute for learning DAX. People at my work often take shortcuts and think they can ask questions without having any context on the subject matter. I guess they are simply too lazy to learn.
I use ChatGPT as a tool to help me, but I believe that learning the subject matter is important. For instance, I have to learn DAX and not rely 100% on ChatGPT.
Agreed. Thank you.
Excelent Video!!! Thanks
You're welcome
Great video, thanks. As a trainer I sometimes have difficulty to explain difficult DAX formulas to trainees. I guess I could use GPT4 to get explanations of self-built formulas.
Definitely. It's great at doing that
Great video, thanks i will try this out
Thank you
thx for your Effort bu the problem here in saudia arabia there is no chatgpt inn saudia arabia
Ah, that’s a shame
Maybe asking for concurrent guests/bookings would work?
Is that just a question Paul or suggesting a change to something in the video
It's amazing for now, but I think we are on the way which machines will replace humans :(
I think it’s initial impact will be to massively improve the efficiency of those that use it.
It will also reduce but not remove the need for many specialist areas, which has been steadily the case for many years now.
I tried it with Google Bart and it came up with the following DAX:
// Create a variable for the reservations table
var reservations = CALCULATETABLE('Reservations');
// Create a variable for the calendar table
var calendar = CALCULATETABLE('Calendar');
// Create a variable for the number of rooms booked on each day
var roomsBooked = CALCULATE(
SUM(reservations[Room]),
// Filter the reservations to only include days that are in the calendar table
WHERE(
reservations[Arrival]
Hah! That's pretty awful
GPT is nowhere near to being as good at DAX as it is with other languages because DAX is typically contained in PBIX files and those are not source code. There are barely any GitHub repos containing DAX code. The model could never learn it properly.
Interesting point. The huge challenge also is it’s so dependent on the data model and structure of underlying tables