·By David
Ai Software Development, who's doing the testing?
#software#R&D#development#IT security
As someone who ran software dev teams over 30 years, and coded myself since the 80's, I must say I am alarmed at how quickly Ai builds software platforms and the fallout that's going to create for developers and companies with 'legacy' software platforms.
I was in IT security during that time and am even more concerned about Ai software systems apparent lack of proper security testing before an application can be published. We are going to see some significant breaches over the next few years with firms heading to Ai without realising what they are signing up for.
Having built a few test platforms I went full steam to design and build a membership and events platform for my car club. We had it live in three months but then spent six bug fixing and some of these bugs were serious holes in the system. It's very hard to get your head around the code Ai is creating(just as it is looking at other devs code) and to hold back and test when that big 'Publish' button is on screen.
I have spent a year with them now and the Ai software development platforms have worked hard to sell 'people with an idea' for a software platform a way to get it created and published as quickly as possible; most of these people will have little to no software programming skills, let alone QA style and security testing.
In the drive to increase profits, for their own bonuses and shareholders dividends, senior management will mandate that software dev teams are reduced in scale and Ai 'plugged in' in their place. They won't listen to the few devs that at risk of being made redundant by Ai dev platforms have actually looked at these systems and tried them out.
If you look at this, yes Google Ai generated, list of companies jumping already we should be concerned, very concerned. I hope some of those saved costs are being redirected to testing.
-----
Major technology and enterprise companies have reduced software engineering headcounts or attributed labor efficiencies to AI integration, though complete 1:1 replacement of human engineering teams with AI is rare. Instead, companies are scaling back traditional coding roles while reallocating budgets toward AI infrastructure and AI-assisted development workflows. [1, 2, 3]
Companies Citing AI Efficiencies or Reducing Developer Staff
- Spotify: Publicly attributed operational and labor efficiencies partly to AI tooling alongside structural adjustments. [1]
- Atlassian: Noted labor efficiencies and shifts in engineering capacity influenced by AI adoption. [1]
- Block: Adjusted team structures and pointed to AI-driven productivity gains. [1]
- HubSpot: Implemented significant workforce reductions (laying off hundreds of employees) while pivoting operations around AI-first workflows. [1]
What is Actually Happening on the Ground
- Augmentation vs. Full Replacement: Most firms are not literally firing all human programmers to run on pure AI. Instead, they expect remaining engineers to use AI tools (like GitHub Copilot or Claude) to do more with smaller teams. [1, 2, 3, 4]
- Shift in Skill Sets: Routine coding, boilerplate generation, and tier-1 support roles face the highest compression, while demand has risen for engineers who can validate AI code, handle security, and focus on system architecture
-----
Gatehouse Way, Aylesbury
Leaflet © OpenStreetMap contributors