Where's the guarantee for recognition of future growth....if they don't recognize past growth?
The biggest gripe I have about articles such as this is that it assumes a static perspective of "now, into the future" and it doesn't account for "all the time before now".
If I'm having a conversation akin to the one that opens the blog post, then presumably I've been at the company for a while. Conversations like that don't just happen between CTO and engineer unless there's some time vested in the company for both.
A CTO saying "take my job" as a non-sequitur is sus, IMHO. Now if it's said in the context of "here's a raise, and if you want another one....try to take my job", well now there's some decent context for the ask and a reason to believe that future growth will be compensated.
The best prediction of future performance is past behavior. That goes for mgmt as well as pee-on.
Now an AI-generated PR summary I fully support. That's a use of the tool I find to be very helpful. Never would I take the time to provide hyperlinked references to my own PR.
I don't need an AI generated PR summary because the AI is unlikely to understand why the changes are being made, and specifically why you took the approach(es) that you did.
I can see the code, I know what changed. Give me the logic behind this change. Tell me what issues you ran into during the implementation and how you solved them. Tell me what other approaches you considered and ruled out.
Just saying "This change un-links frobulation from reticulating splines by doing the following" isn't useful. It's like adding code comments that tell you what the next line does; if I want to know that I'll just read the next line.
But I explained to the AI why we're doing the change. When the AI and I try something and we fail I explain that and it's included in the PR.
The AI has far more energy than I do when it comes to writing PR summaries, I have done it so many times, it's not the main part of my job. I have already provided all the information for a PR, why should I repeat myself? What happened to DRY?
But that's not what a PR summary is best used for. I don't need links to exact files, the Diff/Files tab is a click away and it usually has a nice search feature. The Commits tab is a little bit less helpful, but also already exists. I don't need an AI telling me stuff already at my fingertips.
A good PR summary should be the why of the PR. Not redundantly repeat what changed, give me description of why it changed, what alternatives were tested, what you think the struggles were, what you think the consequences may be, what you expect the next steps to be, etc.
I've never seen an AI generated summary that comes close to answering any of those questions. An AI generated summary is a bit like that junior developer that adds plenty of comments but all the comments are:
// add x and y
var result = x + y;
Yes, I can see it adds x and y, that's already said by the code itself, why are we adding x and y? What's the "result" used for?
I'm going to read the code anyway to review a PR, a summary of what the code already says it does is redundant information to me.
I look forward to the people who always claim “taxation is theft” to comment on a single man deciding to strong arm a company into giving 10% to the government.
Isn't it the opposite of a bailout, given that the US gov't is seizing an ownership stake retroactively based on past grants/bailouts but giving no new money at this time?
CHIPS act is a grant similar to a small business grant, not a bailout at all. It was intended to incentivize chip production in the United States and was available to any company manufacturing in the united states. It had no equity strings attached, as authorized by congress.
> The government’s equity stake will be funded by the remaining $5.7 billion in grants previously awarded, but not yet paid, to Intel under the U.S. CHIPS and Science Act and $3.2 billion awarded to the company as part of the Secure Enclave program. Intel will continue to deliver on its Secure Enclave obligations and reaffirmed its commitment to delivering trusted and secure semiconductors to the U.S. Department of Defense. The $8.9 billion investment is in addition to the $2.2 billion in CHIPS grants Intel has received to date, making for a total investment of $11.1 billion.
So it kinda is something weird? It's not really a pure bail out, the Chips act already did that, and it's also not really a tax because they aren't going to get money out unless there's dividends. It's more like a power play which makes sense given that Trump is uncomfortable without anyone getting anything for nothing.
This is most definitely an invitation to abuse your phone’s battery, but at the same time I absolutely love this idea. It’s hilarious to imagine someone eagerly awaiting the chance to log onto the site as the battery dips from 8, to 7, to 6. “Just a couple more minutes…”
> What I tend to do is write stubbed out code in the design I like, then I'll get an LLM to just fill in the gaps.
This seems like an interesting approach, though to me it begs the question: what does "stubbed out code" look like? How much stubbing is done? Have you considered using pseudocode as comments within a larger "stubbed out" portion?
The importance of rules and context has begun to elevate its significance (...that is, if context wasn't always very important), and finding ways to articulate that context seems to be a skill of greater importance...
There is a track on musicforprogramming.net which has a passage of music punctuated by someone chewing loudly with their mouth open. I was deep in a flow state the first time this track started playing and the utter disgust was, as the title suggest, unbearable. I recommend jettisoning that track from your playlist if you, like I and this author, share this aversion
Interesting read. Time wasn't a variable I had considered missing from interactions with AI, but it makes sense.
I'd also add this: tools like the AI bots so prevalent today are flawed because they cannot consider things like context, limitations, dependencies and scope. I give a question...they attempt to spit out a complete answer with complete disregard for the context which my question is coming from.
AI fails in the same way a monkey can't drive a car.... abstraction. We humans know a red light ahead means stop at the stop light, not stop immediately where you are right now. All AI can do is make a best guess of what the inputs pattern-match to. This is like always having an answer without ever asking for clarification or context.
Exactly. What I consider a patch and definitely a symptomatic solution is "solved" via agents that search the web (e.g. asking for the weather forecast of this year - in that case the LLM cannot know the year I am referring to, if not via a web search).
Generally speaking, LLMs lack direct temporal awareness. Standard models do not model the flow of time unless explicitly. Some models can encode a model of time when trained on sequential video data and rely on external encoders to provide temporal structure. But that is a very narrow application (video in this example). That cannot be considered a generic form of awarenes of time as a concept through which facts can change.
Where's the guarantee for recognition of future growth....if they don't recognize past growth?
The biggest gripe I have about articles such as this is that it assumes a static perspective of "now, into the future" and it doesn't account for "all the time before now".
If I'm having a conversation akin to the one that opens the blog post, then presumably I've been at the company for a while. Conversations like that don't just happen between CTO and engineer unless there's some time vested in the company for both.
A CTO saying "take my job" as a non-sequitur is sus, IMHO. Now if it's said in the context of "here's a raise, and if you want another one....try to take my job", well now there's some decent context for the ask and a reason to believe that future growth will be compensated.
The best prediction of future performance is past behavior. That goes for mgmt as well as pee-on.
reply