What I don’t get is that PCOM was one of the most prominent features that where promoted right from the start and they obviously did not think about designing / building the architecture needed to support that right from the start. Probably a wonderful example of a case where an agile software development approach isn’t the right one to use, even if it’s the process model everyone talks about (way overrated imo).
1 Like
I get the impression that the marketing side of NDSP strongarms the engineering side a lot. The company makes decisions that really make me scratch my head.
3 Likes
Was about to say the same as @JLantz …
Defo feels like over-promising from product/marketing before actually consulting developers, or looking into the effort required.
Either way, not too long now.
1 Like
I think they locked Doug in a gym and took away his internet privileges.
(I miss him)
3 Likes
Where’s the update?
Ask that question on Sunday night.
Machine learning and AI’s are useful not only for accurately emulating tones, as they are used for the QC, but also for programming.
A lot of shops are already using Chat GPT and other AI “engines” to generate large bodies of code that their developers then correct, refine, and enhance. Eventually it will also enable much more rapid and comprehensive QA and beta testing, ultimately with the ability to fix bugs on-the-fly without requiring a programmer’s intervention. The ability to create and test new features and firmware quickly is going to accelerate dramatically over the next 10-15 years. That process has already begun.
1 Like
Interesting comment. My experience has been with problems caused by the opposite situation, where the engineers’ efforts are not being sufficiently influenced and determined by the sales and customer support departments within the company - the front-facing parts of the company that tend to be more tuned in to feedback from the folks using their product.
In that situation, the software engineers start producing features that may be technically impressive and quite an accomplishment, but in no way reflect what users actually want. Just because it is a nifty or elegant piece of code does not ensure it is something that end-users find useful. Processes within the company that keep the architects, designers, and developers well connected to what customers need are a formula for keeping the product and customers in synch.
1 Like
We’ve spent some time looking at these “AI Tools” at my work. Results ranged from “questionable” to downright unusable.
There are times when they can help, but that is massively dependant on what you are trying to do.
I would argue you are still likely to have far more success using a good template or copying from Stack Overflow from what I’ve seen.
5 Likes
I like GitHub Copilot, really got me back into programming after years of boring managers roles. Amazing how good it got in just two years. I don’t see it take over programmers but speeding up their work will definitely happen. Not too fond of GenAI in general btw, we need humans to be smart, not AI.
2 Likes
AI tools are far from a toy, although they are in some respects in their infancy. They are being used to take a lot of the grunt work out of coding and can be enormous timesavers already. There will likely come a time not too far in the future where if you are the developer at your company not using something like Chat GPT to assist in initially generating your code, you might just be the slowest coder on your team.
AI tools are like a blunt force first-take. I have seen countless hours and tedium saved using them on projects. No argument a great programmer can write better code, at least for now. An even better use of their time can be to use, for example, Chat GPT to take a first swipe at the code, or at least parts of it, to be completed and refined by the developer (per my previous comment below).
Great programmers also tend to be better at using all kinds of programming tools including Chat GPT to get superior results. There is a heavy GIGO penalty when using Chat GPT. To produce higher quality output, you must supply optimal and often very specific instructions and constraints. That can require an iterative process to refine the code generated but can eventually produce some very usable results. Even if they require additional editing.
This is still a much faster method for generating a variety of blocks of code than writing them from scratch. Those benefits become even greater when re-using or adapting previous Chat GPT code generation sessions.
I guess time will tell but IMO these tools will be ubiquitous within a short span of time. Given long enough, I believe a good many programmers will have transitioned to meta-programmers - those who have the skills to tell an AI exactly what code they want written and what that code should do. Coding the nitty-gritty will be an archaic exercise and somewhat like bringing an abacus to an advanced mathematics class. Which is not to deny, that on occasion, the abacus might be the best tool for the job. 
AI will improve. And humans, hopefully, learn to use it as intelligence augmentation tools. AI should not be used to replace humans, that could lead to scenarios I don’t even want to consider.
I work, as an AI specialist, for a large international bank and my attempts to slow down the development of useless AI tools doesn’t resonate. Same thing happens in many other companies. Visions of ‘a personal AI bot for every task for every customer’ are abundant and scary. And absolutely unnecessary as we need equal amounts of people that make, change, monitor, edit, audit etc. the AI tools. Crazy!
In case you have not seen this video from Rick Beato, it’s worth checking it out:
2 Likes
He can really rant sometimes, or a bit too much to my taste actually, but he’s spot on with this topic.
Some sobering comments from the “Godfather of AI” in this ‘60 Minutes’ piece.
Along with the potential benefits there are also some very dire possible outcomes. Ranging from massive disinformation campaigns, millions of jobs being eliminated, to SkyNet (Terminator) scenarios at the dystopian post-apocalyptic end of the spectrum. Buy hey, we’ll probably get some awesome modelers out of it before then.
Love Rick Beato’s videos! Great musician but what truly stands out for me is he is one of the most respectful interviewers I have ever seen. Listens more than he talks and that makes for some real revelations from many a top-notch musician.
1 Like
We’re far from on-topic now but what do you rather see? AI generated art that takes a few or less seconds to generate, trained on data from the best that humankind produced or the real thing?
I mean…
versus
I’ve seen Rembrandt’s The Night Watch in person. There is absolutely no comparison.
I prefer the dog and cat
4 Likes
And the last work day of the month has passed and nothing about the update…
3 Likes
Let them cook…
Or as GNR would say:
“Said, woman take it slow It’ll work itself out fine
All we need is just a little patience”
2 Likes