Had been consulting for equity with one startup out of an accelerator, so it was natural to go paid once I went out on my own. For the next few clients, approached investors I knew from that and other startups, who referred me to portfolio companies who needed me. I wish I'd read Alan Weiss's Million Dollar Consulting at the beginning though, I would have avoided many mistakes (like day-rate billing).
The author replied quickly and described his use of AI as very limited and just for grammar and wording. I believe him, based both on the text of the article itself and what he told me.
Thanks for writing that up. You convinced me that there was more Claude here than I'd thought, but I didn't see evidence that the author hadn't edited and supplemented, which is what I was suggesting. In fact, your last observation about correcting an erroneous date makes my point, not yours: Claude made a mistake, and the (human) author fixed it, thus improving the essay.
I certainly agree that the author should disclose the use of an AI, how much is human vs silicon, and clarify which ideas are his own and which are not. I've written to him to ask about that.
The article is well-written and makes cogent points about why we need "centaurs", human/computer hybrids who combine silicon- and carbon-based reasoning.
Interestingly, the text has a number of AI-like writing artifacts, e.g. frequent use of the pattern "The problem isn't X. The problem is Y." Unlike much of the typical slop I see, I read it to the end and found it insightful.
I think that's because the author worked with an AI exactly as he advocates, providing the deep thinking and leaving some of the routine exposition to the bot.
This is not true for business books like mine. It's vital to write a proposal first in that world; publishers want to influence the content (as in the OP article).
I think the same is true for tech books but I don't know as I haven't written one.
A novel or other fiction is the opposite; there you do have to write the whole thing first.
Certainly, we're at the "bash it with a hammer" stage not ready for anything nuanced. I just wouldn't want to assume that the right outcome is "less autism"; I suspect most people could do with at least a little more!
As I commented in another thread, there's no a priori reason to believe that the "average" glutamate receptor level is the "right" one. Isn't it possible that there are:
1. "Normal" people with a level of glutamate receptors at 10, say, on a scale I'm inventing for this example
2. "Autistic" (according to the DSM) people with a level of, say, 5, who are hindered by the effects of being at this level
3. "A little bit autistic" people at a level of, say, 8, who aren't hindered and don't meet the DSM criteria, but in fact actually benefit from the effects of being at this level
Some "normals" might then want to inhibit their glutamate receptors somewhat to get the benefits of being at an 8 or a 9 on my made-up scale.
It seems you are assuming that because the majority of people have a certain quantity of glutamate receptors, that they are the healthy ones and that we should be trying to bring autistic people up to that level. Is that right?
Why not consider the opposite, that the most beneficial quantity of glutamate receptors could be somewhere below the typical amount? If that were true, then we could try to help others reduce their glutamate receptor level to become healthier and more successful (and a little more autistic).
If we found, say, an association between a lower level of neurological characteristic X and concert-level piano skill, then those who aspire to play that instrument at an elite level might try to decrease X. The fact that most of us are rubbish piano players would not be evidence that lower levels of X are harmful, but very much the opposite.
Amazingly, no one seems to have actually checked that this picture was really "circulating on social media". I've been investigating for the past hour or so and can't locate a single public post or reference anywhere other than reposts of the BBC article.
Typically, postings that gain traction have many many reposts and though some may be deleted, there's a long tail of reverberation left behind. I can't find that at all here.
I wonder if the hoaxer just emailed it to Network Rail directly?
Doing it without your customer's agreement is indeed unhelpful! I'm sorry you're that frustrated by this experimentation.
That's why I advocate using feature flags and beta labels and supervised user research to get feedback. Do those methods work for you, so you can opt in?
Thanks! That's not the positive use I have in mind here, where a manager has a real need and uses a demo to focus the work. Does that ever happen at your large corporate?
Roadmap [1] last updated in June 2025, so a reasonable chance that the project is alive. Though the status colours indicate there's a good percentage of development left to do even before early access.
Yes, the incorrect usage is widespread! See Fowler's original book for the thinking behind the term -- every example therein is a 1-minute job, and many are macros in your IDE.
Good point that LLMs tend to rewrite unless corrected. I have heard (but not tested myself!) that if you tell them to apply a series of small changes they stay on track better. Fowler's list would probably be a good starting place.
I have to imagine that like pair programming, this multi-AI approach would be significantly more tiring than one-window, one-programmer coding. Do you have to force yourself to take breaks to keep up your stamina?
A friendly reminder that "refactor" means "make and commit a tiny change in less than a few minutes" (see links below). The OP and many comments here use "refactor" when they actually mean "rewrite".
I hear from my clients (but have not verified myself!) that LLMs perform much better with a series of tiny, atomic changes like Replace Magic Literal, Pull Up Field, and Combine Functions Into Transform.
Creative lawyers will be all over this. First you get the officer to testify that AI helped write the report, then you call the AI as a witness. When the judge tosses that, you start issuing subpoenas to everyone you can find at OpenAI and Axon.
As others point out, the actual bodycam footage will be definitively probative for the events it records. But there are plenty of cases where the report itself leads to later actions that may be tortious or criminal, and finding out who's to blame for the exact wording used is highly relevant.
Example: AI incorrectly reports that during A's arrest, A made incriminating allegations about B. Based on the report, the police get a warrant and search B's house. When it turns out B is innocent, B sues the department, and when the report turns up during discovery, we're off to the circus.