12 minutes ago
Now I understand High School
1 day ago
The AI Cut and Paste Dilemma
3 days ago
The Era of “I think this is the problem.”
4 days ago
Counting Token Shock
5 days ago
Up at 5 am for Saturday Morning Cartoons
6 days ago
I Can do That!
7 days ago
This Looks Like a Good Problem to Solve
1 week ago
No One Person is the Meeting
1 week ago
Worth the Travel?
1 week ago
Training an LLM for a Marathon
12 minutes ago
I was never a math or science major in school, and strongly disliked having to show my work. I wanted to figure out the answer in any way I could, and just give out the answer and get the full marks. Now with AI, I’m seeing code and ticket changes that go from a blank class file to a fully fleshed out solution – that might work well – but hasn’t been removed. It’s a good reminder for me as I use AI, that yes, it’s important to keep showing my work. The differentiator won’t be the answer; we all …
1 day ago
The AI Cut and Paste Dilemma
Greg Thomas
When a ticket goes from no content to 1000 lines of content, here is where my mind immediately goes now. I use AI to help me troubleshoot and track down problems – no one knows it all – and that goes for you to AI! After being burned a few times, with AI usage, I have found myself now in an instant doubting mood when I see long pieces of content come my way, where I now have to remind myself – someone might have put a lot of effort into this and I should give it the time it …
3 days ago
The Era of “I think this is the problem.”
Greg Thomas
We are exiting the era of someone knowing the answer and entering the era where everyone around the room can say – “I think it’s this”. During a call, everyone can bring up their AI tool du jour, copy and paste an error message in, and come back with a “I think it is this”. It’s not the answer; there is no analysis, there is no surety in what we are trying to solve. Having more information is great, but if it’s not directed to the solution at hand, it’s just more noise. Whereas we used to have people who …
Training an LLM for a Marathon
Training a Language Learning Model is much easier than training to run a marathon. You assemble your data, you set your parameters (depending on the scope, you do a bunch of other things), and then you click “Train”. And it runs, for however long it needs to run, it runs. It does its job, you walk away and let it work, let it run, let it train. Everything we do today is not nearly as clean or as easy as clicking the “train” button – if it were, I’d be selecting different models and hitting “train” all the time. Want …
2 weeks ago
Your First Meeting
As far as I can think back, my first professional meeting was when I was working part-time at A & W, and we had to have some sort of staff meeting on a Sunday night. I had no idea what was happening, but I got paid to be there, and we had free pop and danishes. The Impetus for the meeting is someone had just been fired, and it became a bit of a forum to air dirty laundry. My biggest issue was that my pants were 3 sizes larger than I wore, and it was an awkward belt I …
2 weeks ago
One thing AI will never be able to do
Spend 3 hours trying to force your way through a problem because you’re a developer and you think you can code anything. So you spend 3 hours going up against the wall, even using Claude to help you with what you are doing wrong. And then you finally read the KB article associated with the error and realize you just wasted 3 hours. But wow… what a day of trying to make the impossible work. Want more? Check out my book Code Your Way Up – available as an eBook or Paperback on Amazon (CAN and US). I’m also the …