Covenant0028

Exactly. How is it that an org suddenly discovered thousands of employees are under performers? And how is it that the number of under performers coincides with the number McKinsey and Co (or similar company) said it would be?

My theory is that when an organization descends into a cycle of repetitive downsizings, it inevitably leads to people focusing more on protecting their jobs than on business value.

By the process of natural selection, the people who survive such rounds of firings are the ones with the political skills to survive them. Some of those guys will be star performers, but most will simply be politicians. Over time, this process will result in an increase in the ratio of politicians to star performers.

Give them a few months and it will be called Microsoft Copilot AI ML Inferencing Engine for Microsoft Windows and will come in 5 editions.

I suspect the analog will be that the "human in the loop" will bear all the consequences. Perhaps even if they did nothing wrong and are in fact the victim in that situation.

Take the case of Linda Yaccarino. Ordinarily, if a male employee publicly and sexually harassed his female CEO on Twitter, he would (and should) be fired immediately. When Grok did that though, it's the CEO who ended up quitting.

Another thing that would permanently taint models and open their creators to lawsuits is if they were trained on many terabytes worth of pirated ebooks. Yet that didn't seem to stop Meta with Llama[0]. This industry is rife with such cases; OpenAI's CTO famously could not answer a simple question about whether Sora was trained on Youtube data or not. And now it seems they might be trained on video game content [1], which opens up another lawsuit avenue.

The key question from the perspective of the company is not whether there will be lawsuits, but whether the company will get away with it. And so far, the answer seems to be: "yes".

The only exception that is likely is private repos owned by enterprise customer. It's unlikely that GitHub would train LLMs on that, as the customer might walk away if they found out. And Fortune 500 companies have way more legal resources to sue them than random internet activists. But if you are not a paying customer, well, the cliche is that you are the product.

[0]: https://cybernews.com/tech/meta-leeched-82-terabytes-of-pira... [1]: https://techcrunch.com/2024/12/11/it-sure-looks-like-openai-...

Vibe SQLing is where it's at

I've had similar thoughts as well. Rust or C would also be the ideal candidate from an energy consumption point of view, as they consume way less energy than Python, which is the language that LLMs most often default to in my experience.

However it's unlikely LLMs will generate a lot of Rust code. All these LLMs are generating is the most likely next token, based on what they've been trained on. And in their training set, it's likely there is massively more Python and JS code out there than Rust, simply because those are way more popular languages. So with Rust, it's more likely to hallucinate and make mistakes than with Python where the path is much better trodden

Using IP theft to improve a nation's economic and scientific standing is pretty common. The US did it too [0][1]. This is not whataboutery, but a simple fact that industrial espionage was the ladder that today's developed nations used, only to then kick away the ladder when the rest of the world tried to catch up.

0: https://www.smithsonianmag.com/smart-news/how-industrial-esp... 1: https://www.newyorker.com/magazine/2014/06/09/spy-vs-spy-3

It's more likely that the user will need to ask the agent to solve the CAPTCHA, because right now AI bots are better at solving CAPTCHAs than humans are.

> Agents, while leveraging similar systems, are serving a site's end consumer. When I use an agent to shop, I'm still the customer of the shop. As the shop owner, I want to give the best experience therefore it's in my best interest to provide an AX that supports them providing a good experience to the end user.

This is fine until the agent decides to order something the customer did not want. This is inherent to the concept of an agent. Due to the probabilistic nature of LLMs, and the fact that no agent will ever be perfectly able to predict exactly what you want at the time you want, this scenario is inevitable.

As the shop owner, this would result in an increased numbers of returns. You could recommend that the user must approve the purchase, but given that you do not define these agents, there is no way for you to ensure that the user is actually following your advice.

> The only argument to be made in this department is proprietary kernel level anti-cheats

Just last year we saw that an actual cybersec company like Crowdstrike couldn't prevent a massive screw up with kernel level code. Given that, I am stunned that people are willing to trust game publishers with access to their kernels. Even leaving aside abstract notions of freedom, do people not get that this is extremely poor security practice? Is any game worth the risks that come with giving random game companies (who are presumably less cybersec savvy than Crowdstrike) the keys to the kingdom?

This pearl clutching over piracy is extremely ironic in the age of AI. For years it's been said that piracy is akin to stealing money from creators of all sorts. And ostensibly to protect the rights of those creators we've had various laws and tech solutions like DRM in place. Those tech solutions are supplemented by further laws that make it illegal to break DRM even on content that you paid for and own. Apparently preventing people from breaking DRM on stuff they paid for, protects the rights of content creators somehow.

Yet in the midst of all this, AI companies can train their models on any and all content, with zero attribution let alone compensation to the people that created that content, and then rejoice at the number of those people that they'll put of a job. So it's a crime to pirate a movie to watch with your family, but take petabytes worth of data from millions of people to train an AI that you will monetize, and that is perfectly legal.

> At least in London you can get a large set of cultural exposure in a single visit.

So why not use this logic to extract all of The Hague's Mauritshuis (including The Girl With The Pearl Earring) to London? Let's include the most prized artifacts of other European countries too in this "large set of cultural exposure".

I love how an extreme example of one country doing this to their historical artifacts, is used to deny entire swathes of billions of people access to their artifacts. So all colonized countries must be damned by the record of Afghanistan (which itself owes a lot of its instability to meddling by colonial powers), while the British Museum must NOT be damned by it's own record of having hundreds of artifacts defaced and parts sold for scrap: https://edition.cnn.com/2023/12/13/style/artefacts-british-m...

> After all, the British Museum, the main example for restitutions, is located in a global city, given completely free access to its huge collection on display and pays for preservation. The global cultural value it adds is much larger than individual museums all over the would could provide.

Firstly, as others have pointed out, the British Museum is not freely accessible to the vast majority of the world population. The world is larger than WEIRD countries and the richer sections of formerly colonised countries countries

Secondly, it is not the British Museum's decision to make about whether the cultural artifacts of other people is more valuable to "global culture" than it is to the culture it originated from. Let's take an example: India and China both have a population more than double that of the EU. Which means a museum located in either country provides visa-free access to billions of people with relatively cheap travel.

Would that be an argument to move the Girl With the Pearl Earring from The Hague's Mauritshuis to New Delhi or Beijing? Would it be an argument to move the painting to the British Museum? Indeed why not move the entire Mauritshuis to the British Museum? After all London receives far more visitors than The Hague does. Surely the global cultural value of those artifacts is greater than the value that the Dutch place on it.

Consider that you apply this logic to extract all Dutch artwork from that country and place it in a second country. Consider that the second country was largely responsible for such extraction, which included not only cultural artifacts but also wealth. Consider that the second country now places visa restrictions that make it harder for Dutch people to visit the country and even if they did, the cost of actually doing so would largely exclude most Dutch people. What effect would this extraction of cultural heritage have on the Dutch?

> Otherwise, unless there is a clear claim of individual/familial ownership vis descendancy, then returning artifacts to a legal national government is still a least bad option.

How do you expect such a clean and clear historic record considering these artifacts were taken from places subjected to deep exploitation by the same countries that today possess the artifacts? Surely keeping records of historical artifacts for use centuries later, would have been a lower priority for the local population than survival.

I would recommend reading "Africa is Not a Country" that delves into this exact question. It is the pinnacle of self-serving arguments for European countries to first spend centuries colonising Africa, stolen its wealth and cultural artifacts, enriching themselves in the bargain, and dividing the continent in a way that would guarantee political instability, to then use that same instability as an excuse for not returning those cultural artifacts.

Of course this article is about Indonesia - that's not a country that screams instability in any case

There is a difference, and AI companies understand it very well. All of them prohibit you from using their model to train other AI models. Microsoft takes it a step further and even prohibits you from trying to discover how the models work.

No human, however powerful, can prevent you from looking at their actions and learning from them. You can look at Obama's speeches for instance and learn how to craft certain messages for your own speeches. Nothing he can do to stop you from doing that.

And that is the key difference: AI models have been designed to privatize the process of learning, wherein they have unlimited freedom to learn from any human's work without compensating them from it, but humans or even other AI models cannot learn from an AI model.

This distinction IMO removes any right that the AI companies have to pretend that their models are people. They're not, the actions of the AI companies themselves show that.

Social interaction is not synonymous with the office. Adults can choose to socially interact with any group (family, friends, shared interests) that they want to.

In theory kids can do the same too, but in practice, you'd need some set of adults to get all those kids into the same place somehow, another set of adults to supervise them so they are safe etc... at which point it becomes more efficient to combine this with schooling.

Also with working adults, there is an explicit expectation of delivery. Both the employer and the employee knows of the consequences for performance or non-performance, and part of being an adult is that you are aware of those consequences and choose to behave accordingly. With kids, such expectations aren't realistic.

The idea that individual source snippets at this scale diminish in aggregation, is undercut by the fact that OpenAI and MSFT are both selling enterprise-flavoured versions of GPT, and the one thing they promise is that enterprise data will not be used to further train GPT.

That is a fear for companies because the individual source snippets and the knowledge "learned" from them is seen as a competitive advantage of which the sources are an integral part - and I think this is a fair point from their side. However then the exact same argument should apply in favour of paying the artists, writers, coders etc whose work has been used to train these models.

So it sounds like they are trying to have their cake and eat it too.