AI is enabling low talent men to succeed in the workplace in a way they couldn’t have done on their own.

[lacklustre]

Last week I was in a senior leaders meeting, a man who is knew to the company gave a presentation and hadn’t even bothered to disguise the telltale AI giveaways - the capitalised headings, the emdashes, the lot. The other men in the room praised him enthusiastically for his excellent work and he happily accepted their praise.

I work in a male dominated industry and I’m seeing this more and more and while I love AI and use it as a tool, I’m seeing so much stuff sent to me by male colleagues that’s clearly not their own work. I have a boss who’s decided he no longer needs the services of a consultant because he can be his own consultant with AI. He also runs most of my work through AI and then sends me ‘his’ improved version. Another male colleague even uses AI to write his WhatsApp messages. And they all constantly congratulate each other on their incredible work.

Is this a thing now? Of course I imagine lots of women use it in this way too. But in my large company, this is increasingly common and it’s like robots praising robots. Find it all quite disturbing especially when I know that two years ago, some of them were on work performance plans almost at the point of being managed out of the company.

Does it even matter? If I can’t beat them, should I just join them and start doing it myself?

The field of internet trolling apparently

I agree. I have a fair mix of male/female clients and would say they both use AI just as much as each other, annoyingly often sending me AI generated emails which are just so obviously AI.

Fully agree!

"I gave ChatGPT a complete website today - written originally for PHP 5.6 and said update it to php 8.5 compatible - within 2 minutes I had a zip file - uploaded it and job done - no issues... far faster than I could have managed!"

No issues... so far.

I know nothing about PHP, so don't know how hard that is. So correct me if what I'm about to say is wrong, but my spider sense is tingling. If they are different versions of the same language, PHP, then hopefully it's a straight forward translation.

Any testing?
If you have tests then how good is your test coverage?
How can you be sure no bugs were introduced?
😱

Is it hard - no, not if you are logic driven - but there are different commands / different structures to how you get information in and out of a database etc.

think of it like translating from one language to another - but without the nuances languages bring - so relatively easy... it is a perfect job for AI...

Where AI goes wrong is when it tries to think - and wanders off aimlessly down corridors of irrelevance...

give it strict parameters and tasks of a black and white nature and it does well

I get the idea of what happened. I just asked about testing the outcome to see how well it turned out. Did you compare the new code with the old? Did you test each change thoroughly to make sure the new behaviour is that same as (or better than) what it did before? Basically how can you be certain no bugs were introduced without checking, or do you just trust AI?

I'm very interested. I know people who tell me that an old-style ASP website can't just be automatically translated into newer technology because one can't trust AI to do the translation perfectly.

It was a very simple site - so very easy to validate that it worked - and I had coded the site and have 40+ years experience of coding - so quite easy to visually run through the coding to check - and then test it...

a more complex site (some to come) would I think need a more structured testing program, and more need to revert and correct and refresh as AI can randomly do things...

but in principle the conversion is not too hard - it is though simply one version of php to another which means things like replacing mysql_query with pdo etc. a lot remains the same...

ASP is not a world I play in particularly, so am not so knowledgeable about what it would do, but in theory it should be able to read the code and build a process document from it - check that ad then feed it back in for the language of your choice - there would undoubtedly need to be human intervention - but it is still a huge time saver

for me it is the equivalent of a member of staff who is competent but not excellent - where I check the work carefully, but when given clear instruction gets on and produces good output... and all for £20 per month!

Interesting. For the legacy system I'm thinking about, tests would be an absolute requirement to mitigate risk (it is very complicated, and business critical, which is why no one has had the time/budget/bravery to re-write it - and no one is trusting AI.... so far...).

I can see how converting from any computer language to any another should be theoretically possible because both languages have very strict syntax and well defined unambiguous behaviours. Perhaps the difference between the ASP translation and your PHP translation in the "depth" (to borrow "depth" in a linguistic sense from Chomsky deep vs surface structure)....?

Would I be correct to characterize what AI did was a like-for-like replacement of one set of syntactic forms with another? That this works because sections of code that use the mysql can be replaced one-to-one by sections of code that use the pdo (which still uses the same SQL)?

I don't understand ASP. I'm told that the conversion from Classic ASP to ASP .NET is complex because those two frameworks don't work in the same way, so it is not as simple as a one-to-one correspondence in function.

It is a thing now, yes. I really can't imagine it's a thing confined solely to the males of our species though.

that is a fair analysis...

but every bit of software comes down to conceptual processes - click this button, and this happens etc. - so that can always work as a mid-step in effectively translating language - and arguably you don't need to require a strict match in terms of one language for the other, as long as data / data links / processes / functionality / outcome etc. all remain - and few of those have a specific link to code - so any code which meets the same needs is sufficient.

Thing is, unless you're writing a bespoke marketing piece it's going to mostly just be the same old business waffle anyway. It's highly unlikely anybody is really going to care as long as it relays the requisite info, which is usually that you looked at xyz, found xyz, which means xyz, and you're going to do xyz.

It's like how using a template specifically written for something is usually going to be much quicker than starting from scratch. The key is just knowing how to tweak it to not look like a you've used a template, which is where people usually fall down. But most management reports, presentations, team updates, etc, are not usually exercises where people expect much innovation or flair. They just need to not be confusing or filled with particularly egregious grammatical errors like should ofs/could ofs or mixing up they're/their etc.