I am only reporting what I have heard from people who are using this technology at a very high level. It’s not messaging. Two years ago my friend was pretty dismissive about the impacts of AI, now she isn’t. Two years ago she’d find things that were wrong with the code, now that is a rarity. That doesn’t mean there won’t be tech jobs, but there may not be as many and they might not be nearly as well paid. How long this all takes, of course, is a big unanswered question. And, of course, it is going to affect far more than tech.
3 Likes
I think this is going to be incorrect. Using AI efficiently and well will likely be a highly compensated skill.
1 Like
I agree with your points (and it sounds like we’re both in similar professional situations).
Software engineering and technology development have always been about much more than just writing code. AI tools will make us far more efficient and productive, and they will certainly change how our teams work (and yes, may reduce the need for as many pure code writers), but I don’t see them eliminating the need for technology teams altogether.
7 Likes
I agree—but.
The problem is that transition periods like this, while good for everyone in the long (and maybe even medium) term, create very real pain and suffering right now.
6 Likes
Sidebar for those who love “rankings”: Recent report on university ranking for coding skills 2026 University Ranking Report | CodeSignal
2 Likes
I have two former students. A graduated from college two years ago and is working in a consulting firm, and B is wrapping up junior year in college preparing for summer internship. A’s daily job has a good portion of searching for information, validating them, converting to usable data, and run some basic analysis. B is learning something close to AI engineering, and his summer internship involves training AI to do what A does on daily basis. They know each other, and are joking that B is training computers to take over A’s job. A is also utilizing AI regularly to gather information; however, it’s still necessary to have a human to validate and take responsibilities. It is true that A is capable of taking on more work by utilizing AI, also true that A’s company has laid off some analysts, but it won’t be AI take-over. A needs to spend significant amount of time checking what AI provides, and once a while when something goes wrong from what AI provided, a lot of time is needed to find out what went wrong where and how.
1 Like
I have a colleague with an anxiety disorder who reported having a full blown attack over the weekend. Could not get hold of her shrink, was current on her meds– did not know what to do. Her spouse logged on to one of the new AI mental health provider chatrooms (I don’t know what it’s called), handed spouse the laptop and said “let’s try this”.
On Monday, after feeling MUCH better since doing the AI session, colleague got hold of the regular shrink. Who basically said the exact same things that AI did, walked her through the identical grounding exercise, suggested the same post-episode recommendations. Offered to up the meds (which AI had not offered, since AI is not a board certified physician although you have the option to do a telemedicine session with one as an add-on).
My colleagues conclusion- AI for the win!!! Clearly not suitable (yet) for someone who is having an active psychotic episode who needs a hospital, a diagnosis, medication, and in-patient follow up. But for the “rank and file” who can’t get in touch with their regular provider- WOW.
2 Likes
Interesting.
“Unlike traditional university rankings that rely on reputation surveys and funding metrics, our ranking is based entirely on the demonstrated programming and software engineering skills of each university’s students, as measured through standardized assessments.”
I wonder how the samples were selected and what standardized assessments were used.
1 Like
Looks like that company (codesignal) provides pre-hiring assessment tools ( and other pre-hiring components). Seems that they could easily look at the performance on these assessments and correlate them with the schools the applicants attended.
and here come’s Meta with a bonkers but classic Zuck idea.
1 Like
They need to get out more.
The keyboard analysis game has been going on for at least ten years (maybe more) in financial services and fintech. Any company whose business could allow an employee to transfer funds to their own off-shore, untraceable account (or Bitcoin) with a click has had some form of keyboard analytics for a long time.
I have a friend whose company was tracking excessively long trips to the bathroom. Technically, they were just tracking how long a keyboard was going silent during a work shift. But the end result was employees having to explain their…. “functioning” or cigarette breaks.
Using AI instead of just a sophisticated algorithm isn’t nearly as groundbreaking as these folks seem to think!!!
3 Likes
And, the great coders today are training AI well to take over their job. It is happening….was starting at the low entry coding level but it’s getting smarter every day that a high level coder teaches it something. The good thing is, you still need to know and understand the business you are integrating AI into to make it work right, at least for a little while longer.
Time to find a new workplace.
4 Likes
There is a huge variation among different companies, and media has a history of making things sound more extreme than the reality, as doing so results in more clicks/views/profit for the media company. I also work in tech. My personal experience is nobody at the company where I work is being told to use AI. Reasons for this both include being smaller, using older code, and results being dependent on things that require more human intervention such as physical hardware/setup instead of just dependent on software code.
Many individual engineers I know have attempted to incorporate AI or tried using AI, with varying degrees of success. A common complaint is a long the lines you mention, with code that looks proficient to a non-expert, but an expert would recognize as having numerous issues. Or making design mistakes would seem obvious to human, such as verifying code by using a 0 input → 0 output test case. There is some potential there, but requires an expert to inspect and verify anything written by AI. It’s usually more time efficient to modify/update code by hand instead of directly using AI. The potential benefit I see is when there are large volumes of updates – instantly updating millions of lines of code in a minuscule fraction of the time it would take to do by hand. However, one needs to have a supporting infrastructure in place to make this possible that many companies including my own do not have. It also needs to be a company that would frequently want to make large volume code changes, rather than make minor updates to existing, which again does not align well with my company.
The potential impact on new grads is complicated. New grads are often at the forefront of using and understanding the newest technological changes. This likely includes knowing how to code while utilizing AI and AI tools to increase productivity. It can be easier and less expensive to hire new grads who know how to use AI tools effectively than re-train experienced engineers to dramatically change what they have been doing for more than decade. However, new grads are not sufficient for “requiring an expert to inspect and verify anything written by AI”, which has been mentioned as a concern at my company. If a new grad likes using AI to write their code for them, they may not be experienced enough to understand when the code is not having the desired result.
As I’ve noted in previous posts, as a whole post grad outcome surveys do not show dramatic changes – 2025 CS outcomes look much like 2024 outcomes. Of course there are individual colleges that are outliers (either positive or negative) for various reasons, but as a whole, the overwhelming majority of CS grads still are finding jobs, and those jobs average higher early career earnings than all other majors with sufficient sample size.
A few more outcomes I found with a quick Google search are below. Class of 2025 outcome stats for numerous other colleges are also available.
WPI
2024: CS (255 grads, 96% knowledge rate): 16% seeking, median salary = $90k
2025: CS (270 grads, 95% knowledge rate): 9% seeking, median salary = $90k
2024: All Majors (96% knowledge rate): 10% seeking, median salary = $80k
2025: All Majors (97% knowledge rate): 12% seeking, median salary = $80k
U Arkansas
2024: CS (135 grads, 83% knowledge rate): 17% seeking, median salary = $80k
2025: CS (199 grads, 78% knowledge rate): 14% seeking, median salary = $83k
2024: All Majors (66% knowledge rate): 8% seeking, median salary = $60k
2025: All Majors (72% knowledge rate): 6% seeking, median salary = $62k
U Oklahoma
2024: CS (83 grads, 82% knowledge rate): 21% seeking, average salary = $84k
2025: CS (90 grads, 78% knowledge rate): 9% seeking, average salary = $86k
2024: All Majors (85% knowledge rate): 15% seeking, average = $60k
2025: All Majors (83% knowledge rate): 21% seeking, average = $63k
Baylor
2024: CS (21 grads, 100% knowledge rate): 10% seeking, average salary = $91k
2025: CS (32 grads, 100% knowledge rate): 6% seeking, average salary = $94k
2024: All Majors (100% knowledge rate): 7% seeking, average salary = $63k
2025: All Majors (99% knowledge rate): 7% seeking, average salary = $62k
3 Likes
I would also be curious how many of the applicants had a degree in a related field or if there were applicants without a related degree who were just trying their hand at the assessment.
1 Like
Correct. Providing coding assessments to tech hiring managers is a major part of their business, and one that will start to shrink as assessing a candidate’s ability to write code becomes less important for many hiring managers.
2 Likes
My cynical social scientist conclusion: Survivorship bias for the win!!!
4 Likes
Or… placebo for the win!
1 Like
Well, C23 (graduating college this year) had her so far one and only interview this morning. C23’s report is that she basically “blacked out” during it, and so has no report on how she did or even what was covered—but does remember that if she’s a finalist, she’ll be flown in(!) for an on-site interview.
Seems a bit intense for what really is a basic entry-level position, you know? But I think it does fit in well with the level of caution on the part of employers that’s being widely reported lately.
6 Likes
some companies fly in the candidates for an overnight stay and 3-4 rounds of on-site interviews for internships. They may invite a dozen for 2 openings.
2 Likes