ygouzerh

That's a good point.

> Wasting your time... Forcing you to do their work for them

The one forced to do their work for them, are the teammates. Not the manager. The manager only care of it to be done. How it is done: doesn't matter.

I would argue that one of the main reason for meat proxy, is security boundaries.

In the majority of companies with a certain scale, no one can run an AI agent that will do the project from A to Z. You always need to raise at one point ticket to someone else, a human, it cannot even be done by an automated system (technically it could of course, but acceptable by governance... unfortunately no) .

Current companies are paying way more than 200$ per month per developers. In startups, SMEs and big tech, you can multiply it by x10

- job loss and entry level hiring collapse --> I will out that as in the group "Not a bubble" (Not that I am thinking it myself that it's not a bubble).

It looks like the end goal of replacing labor by capital is working.

Interesting, I sent the same wine testing example to Claude Opus 5.5 medium, and it failed, where Inkling 2 Small succeed.

Are they doing that because they are scared that Moonshot might route queries to Claude?

If you succeed to read it (the admin writing style of NIST is quite dense), it's quite good.

It's a bit procedural and admin focus... but if you execute all of it, at least it's easier to defend yourself during an hack towards your clients or third parties.

I would have prefer something a bit more technical, but it's a good entrypoint, at least to help hire or consult a security professional.

+1. The issue is not that us SRE doesn't want to upskill, on the contrary. It's mostly that we doesn't have time, between the backlog always full during the day, and the incident at night. We are considered cost center, and companies try to reduce it a maximum by over-optimizing headcount/tasks

True. In capitalism, the rule is that more capital you have, more capital you will get next year.

On one side it's interesting, because it means that if you buy a machine, next years you can buy more, and more and more things get automated.

The bad side, is that we doesn't have a way for labor to reap the benefits of it as well, except from lower purchasing costs... which is counter-balanced by the need of the system to push for more and more consumption every year, to keep it working. (E.g price of electronics/performance decreasing, but need every few years to buy new phones with more performance to keep up with the more and more hungry apps)

I wished, unfortunately the companies are going to complain that doing so will stop them to have enough money to innovate compared to other countries, or that they are too small to compete. (At least it's the current speech of the lobbies in EU for example).

Never put it resolved

I thought it was the same for here (Hong Kong), but after talking with recruiters and HRs: most of them are just people using bots, and they never read the job description. Very few of these big numbers are actually real candidates that have a chance of getting the role.

They will probably change their name soon

As an SRE, I am seeing the following use case:

We keep struggling with disk capacity on servers being filled, which will then need cleaning. Currently, we got an alert when it's reaching 80%. It's reactive, we need to work on an unplanned task, and it disrupt our sprint and delivery commitments.

Instead, we could pipe the disk capacity data points to Timestral, and we could approximate the date when we will need to work on it, way earlier, so that we can plan our work accordingly.

I guess it might be integrated later directly in tools like Datadog, or Cloudflare straight away, rather than being used by corporates as it is.

After working with AI at work for quite some time... I can envision it. But mostly based on an Idiocracy society than a a Terminator or Matrix like scenario. When LLM is so good, why bother to learn more, when we know we cannot be as good?

And if LLM give us already what we need... why ask her for more complicated things? And so why would the AI improve or be improved?

- 1% of the world population owns 95% of the total world wealth, as per Oxfam - World Food Program from the UN calculated a need to 93 billion per year to solve food hunger in 5 years - A single SpaceX IPO already raised 85$ billion of cash

I would say... a lot.

After indeed, maybe without capitalism we would have not got this amount of wealth already, and nothing stops us to build socialism inside a capitalistic system, by using more cooperative, governement venture capitals, sovereign funds.

The advantage of capitalism is that more capital generate more capital. More machines you have to produce things... which will help you to produce even more machines...

The machines are operated by the labor forces, the real humans. However, the exponential compounding benefits of theses feedbacks loop us are mostly redirected to the ones owning the machines, the owner of the means of production.

Of course, the labor forces operating the machines/means of production benefits it, because they have either cheaper, or more goods. However they doesn't get this compounding effect.

AI is the god of the machines, the ultimate means of production. The amount of labor forces needed to operate it is very few. When the owner of capital invest a specific amount of money for their company, they get X amount of tokens, enabling them to work on Y amount of tasks. It's reward with low-risk.

The drift between capital owners and labor forces will become wider and wider. Until the labor forces doesn't have any more capacity of buying the goods produced. And then the system will implode.

If it was not for a state actor just trying to get information, but for personal gain: he could sell the information to some unethical hedge fund for some good money, ahead of the IPO.

I found an example online, I cannot find the github again, but it was interesting: an instant dynamic generation of React component, you type what you want to build using natural language, and it directly add one to the page.

If I remember well, in the background, it does the multi-choice question: "What is the component requested ? --> Calendar, Slider, ..."

I actually was thinking on the following example, from my personal life: I have two masters, however I was doing summer jobs in my local town, as handyman, and was humbled very quickly.

The others colleagues all stopped school very early, got the same irrational political views that the ones you mentioned, etc.

However, they were beating us totally, students coming from fancy universities, for everything else related to machinery, gardening, repair, etc... Even on new problems, they were able to quickly make these logical connections, that we would miss.

I feel like the ability of making connections is relative to the domain, and more learned by exposure rather than innate.

For example, someone doing system engineering will be able to understand how the networking of a system works, while someone being a movie critics will find the hidden meanings and connections in the movies they watch. Similar for a musician.

Probably indeed the semantic is important. Someone can say something stupid in a specific occasion or all the time in a specific domain, but almost every time there are areas of life where they will be considered "smart".

Cam it be replaced by a Jev-like model? That might be better for latency sensitive tasks

I feel as well that it's the strongest reason. Even without AI, time pressure statistically already increase the error rate

I think here it's mostly that for normal business cases, we doesn't need to build one.

As a DevOps Engineer, I never once saw before the advantage of using a classifier. Now I see multiple parts of the stack where a better level of expressiveness will be useful (PR validations, Blue/Green validation, notification router for alerts, quick smoke tests, etc).

Nobody will give us the time and budget to build a custom classifier for these use cases, but a simple API call yes.

What are you using Haiku for?

Can it be that now they are getting optimized against benchmarks that are valuing logics, rather than human appreciation? (I am not an expert at all, just an idea)

Nice, thanks a lot, I didn't knew about this feature!

I think this is more a quote for engineers

One thing I am missing to be able to move out of Claude Code, is the auto mode (and the soft_deny and hard_deny settings that can be tuned), with it's classifier checking the output.

It's the killer feature from me personally, often when wanting to troubleshoot for example things like Kubernetes workloads. LLMs are now really good at it, but we doesn't want them to like delete a pod.

Other harnesses like Codex have often on static rules, like the allow/deny of claude code, that can filter out based on regex. It's quite good already, but sometimes the model can find a way to write something that wasn't anticipated, or in a convoluted way.

After, I guess it's something that can be added in an open-source harness like Pi, and add like this new Jev model or something else equivalent

Often politics, lack of vision, people being shuffled around, etc (e.g the stakeholder that decided of the acquisition just got moved to an another department, this kind of things)