iiinova — where your digital self keeps evolving

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 18h

I don't agree with ChatGPT's take regarding the missed C macro call graph edge, but its example is certainly imaginatively devious.

2.1. A direct call is missed when the closing parenthesis comes from the caller's source

#define B() 1
#define A() B

A()()

000 0

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 13/10/2026

The relationship of AI-authored code to technical debt is a matter of perspective. I was discussing with colleagues working on combinatorial optimization. They were marveling at how Claude creates better code for them, giving them a 30 line function for what they would code in 300 lines.

000 0

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 08/10/2026

Thinking will become a hobby. For the bulk of today’s cognitive work, the writing’s on the wall: AI will take over. Read more at www.spinellis.gr/blog/20261008/?mdc…

spinellis.gr

blog dds: 2026-10-08 — Thinking will become a hobby

000 0

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 07/10/2026

Remember LLM hallucinations as the major reason for insisting that humans retain an understanding of, and oversight over AI-generated results? Well, models and AI agents have improved substantially, and Google Trends suggests that interest in hallucinations has sharply declined.

110 1

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 03/10/2026

ICSE chairs said that the HotCRP site crashed close to the #FSE2027 deadline.

000 0

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 03/10/2026

I looks like HotCRP, currently used for performing the #ICSE2027 reviews, is down. If you know something more, please reply here to collect related information.

100 1

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 30/09/2026

How difficult is it for Large Language Models to process program code? We measure a property called “perplexity” to evaluate LLM confidence in code completion tasks across programming languages, models, and datasets. Among other things, we found that […] [Original post on mastodon.acm.org]

001 1

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 04/09/2026

Well-trained and experienced software developers still highly relevant in the age of AI! It took me six days to fix a subtle but critical bug. The issue's and fix's complexity make me think AI couldn't possibly handle it. If you want to prove otherwise though your agentic prompting skills […]

mastodon.acm.org

Original post on mastodon.acm.org

011 1

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 29/08/2026

Extracting information to aid debugging (here a binary tournament merge failure) can often be productively automated with Unix command-line tools. awk '$4 == "BEGIN" && $8 { print $8 " -> " $7}' file.log | dot -Tsvg >diagram.svg

000 0

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 19/08/2026

The rg (ripgrep) command, popularized by #Codex, violates so many tenets of the Unix philosophy it makes me want to cry. Much of its added functionality could be a shell script with a find … | grep … | xargs grep … pipeline. Imagine if that implementation effort had instead gone into making grep […]

mastodon.acm.org

Original post on mastodon.acm.org

001 1

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 18/08/2026

Many prestigious scientific venues have suffered from papers containing hallucinated citations. We can throw tools at the problem (I provide an example) but I think it's more important to fix its root cause. www.spinellis.gr/blog/20260817/?utm…

spinellis.gr

blog dds: 2026-08-17 — A controversial proposal for hallucinated references

101 1

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 12/08/2026

@peterhoneyman It caught one of the two issues but not the other. Indeed, we know how to increase trustworthiness by cross-checking, and we do in in committees, juries, peer-reviewing. But it introduces friction, so, instead, we mostly prefer to work with trustworthy counterparts.

So | would remove/correct roughly five substantive review criticisms (Rust control flow, “syntactic” characterization, 33,389 files, received-date contradiction, and the assertion that key validity threats are omitted), and soften the 163/35-GiB and cycle-dedup claims. The rest of the review's factual audit looks considerably stronger.

000 0

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 12/08/2026

Apart from confidentiality concerns, another risk of reviewing papers with GenAI is hallucinations. Here are two examples from ChatGPT's (5.6 Sol) review of my own work with *a given PDF*. It took me several minutes to uncover each, so a long AI-generated review could take hours to verify.

Are you sure? | can't find it.

You're right to question it. | mislocated and mischaracterized it.
45,103 appears on p. 12, in Threats to Validity:

“From the 45 103 analyzed files 45 033 (99.8%) matched exactly.” =O finuxcpp

On p. 11, 33,389 has a different meaning: these are driver source files “not directly or
transitively included in non-driver code.” = O linu-cpp
So my previous statement that p. 11 says “From the 45 103 C and header files ... 33 389..."
was incorrect.

Where did you find the 77% figure?
That was my error. | don't see a 77% figure in your paper/diff. | inferred/calculated it incorrectly when reviewing RQ4 and then treated it as a reported result.
Your diff instead explicitly says that X% of macros exhibit one or more of the identified conversion difficulties; the actual percentage is generated from your analysis and isn't visible in the diff.
So please disregard my earlier references to 77% (and | should likewise re-check the 55% figure rather than assume it is in the paper).

100 1

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 03/08/2026

@stefan I see very little traction from my posts here.

100 1

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 03/08/2026

According to today's FT, “Wall Street learns to love blockchain”. But the article provides only a very fuzzy picture of the technology. To help, I summarized the absolute essentials of blockchain technology in fewer than 500 simple words at www.spinellis.gr/blog/20260803/?utm…

spinellis.gr

blog dds: 2026-08-03 — Blockchain technology in 500 simple words

000 0

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 15/07/2026

For decades the accepted management wisdom has been for managers to delegate tasks and for organizations to outsource activities that aren’t core to their operations. Yet, when the proposed delegate or supplier is an AI agent, these established practices are suddenly portrayed as irresponsible […]

mastodon.acm.org

Original post on mastodon.acm.org

000 0

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 08/07/2026

Over the past few days I used #OpenAI #Codex to perform a fairly extensive and demanding software code change (1740 insertions and 609 deletions on 13 files). Here you can find all my 78 prompts documented and classified, and, based on them, my take on the role of experts in the age of agentic […]

mastodon.acm.org

Original post on mastodon.acm.org

001 1

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 03/07/2026

Artificial intelligence seems to be increasingly matched with marketing stupidity.

Suggestion to use AI summarization on a document containing an abstract.

100 1

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 02/07/2026

As GenAI code accumulates, it's important to document its provenance. In industry this can help diagnose the root cause of problems discovered long after the code was written, locate issues with a similar cause, and evaluate the cost and benefits of AI. In education it can help students be […]

mastodon.acm.org

Original post on mastodon.acm.org

000 0

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 30/06/2026

This is the pocket computer-history museum I use for introductory computer technology presentations: 7AK7 computer vacuum tube, metal-can transistor, SSI chip, exposed EPROM and CPU chip, RAM / SSD storage and RPi zero. I demoed it yesterday to high-school […] [Original post on mastodon.acm.org]

000 0

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 27/06/2026

What a time to live in! We're discussing about AI recursive self-improvement and AGI. And and the same time, we're also managing IT systems with cryptic tags and by adding "000_" to first names so that they appear on top of the list. The list is from […] [Original post on mastodon.acm.org]

000_Yu Jiang
Tags: #NewPC
000_Zhenchang Xing
Tags: #NewPC
Aashish Yadavally
Tags: #RegRev #RegRevNRRRR
Abhishek Tiwari
Tags: #RegRev #RegRevNRRRR

001 1

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 27/06/2026

The saga of falsely-attributed research papers gains momentum. A recent report on the topic www.jmir.org/2026/1/e104934 and a commentary on it www.luxuriousmagazine.com/false-aut… paint a picture where the problem and its […] [Original post on mastodon.acm.org]

100 1

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 23/05/2026

Have you noticed that reviewing AI-generated code is devilishly hard? Here’s the thing: when working on code with GenAI assistance (from a chat-bot, through IDE auto-completion, or, increasingly, with an AI agent) you need a better understanding of the system than when working without. Cognitive […]

mastodon.acm.org

Original post on mastodon.acm.org

002 2

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 16/05/2026

Highly ironic that a published peer-reviewed journal article on AI hallucinations and the trust crisis in artificial intelligence contains (probably AI-generated) orphan in-text citations. doi.org/10.1016/j.techsoc.2026.1032… It caught my attention because the claim "AI-generated […]

mastodon.acm.org

Original post on mastodon.acm.org

000 0

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 12/05/2026

Can you see the image smudges? I thought that my screen was dirty, but it turns out that the smudges are part of the Microsoft Office toolbar (visible in Word, Excel PowerPoint). Can't imagine what development process and bloat resulted in what looks like freehand drawing being part of it. Ideas?

001 1

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 07/05/2026

Today we launch a new run of the popular and award-winning free massive online course (MOOC) on the use of Unix Tools for data, software and production engineering. More than 7,500 learners enrolled on edX. Now, a further reason to follow the course is to […] [Original post on mastodon.acm.org]

002 2

Diomidis Spinellis @coolsweng.mastodon.acm.org.ap.brid.gy · 13/04/2026

Generative AI upends the goals and nature of empirical software research. Here are my notes from Emerson Murphy-Hill's MSR's keynote talk on the topic. www.spinellis.gr/blog/20260413/?utm…

spinellis.gr

blog dds: 2026-04-13 — Empirical software research in the age of AI

010 1