AI/LLMs Plagiarism vs. Inspiration

[gban007]

There is no "accounting shell game" because this is not from their filings: Dario just said, on a podcast, that if their current models were stand-alone products, which they are not, those products would be profitable. But each new generation requires a 10x investment over the previous generations, and those R&D numbers dwarf revenue numbers.

He was not trying to make an argument that the company itself is somehow profitable.

If I have a profitable hamburger stand, so I decide to build another 10 hamburger stands, I'm going to be losing money for a while. But my original hamburger stand, by itself, could still be turning a profit.

I don't know what else to say, and this argument is making me dumber, so I'm done.

But the question that arises is, if they are profitable as stand alone models, how long for? Unlike the hamburger stand, if other companies are investing into the next generation, the current generation will quickly become obsolete and no longer profitable.

So rather than saying LLM 3.0 is profitable, but we are losing money because of investment in LLM 4.0, it is more their overall LLM part pf the business is losing money, as to be able to stay in market / not become obsolete they need to invest more than the profits they are making.

To torture the hamburger analogy, it is like the current stand is making $100k profit a year, but will have to close next year unless spend $1m to keep it up to code to be able to continue to operate. Next year it may somehow make $2m profit, but to keep up to code and not close at end of that year, they will have to pay $10m.

At what point will they be able to stop having to invest so much, how long will it take tk pay off the investments to date etc, to be able to say at end that the hamburger stand was profitable.

 

[Bill Zebub]

But the question that arises is, if they are profitable as stand alone models, how long for? Unlike the hamburger stand, if other companies are investing into the next generation, the current generation will quickly become obsolete and no longer profitable.

So rather than saying LLM 3.0 is profitable, but we are losing money because of investment in LLM 4.0, it is more their overall LLM part pf the business is losing money, as to be able to stay in market / not become obsolete they need to invest more than the profits they are making.

To torture the hamburger analogy, it is like the current stand is making $100k profit a year, but will have to close next year unless spend $1m to keep it up to code to be able to continue to operate. Next year it may somehow make $2m profit, but to keep up to code and not close at end of that year, they will have to pay $10m.

At what point will they be able to stop having to invest so much, how long will it take tk pay off the investments to date etc, to be able to say at end that the hamburger stand was profitable.

Yeah, that’s why they have no choice but to invest 10x in the next generation. That’s exactly what Dario was explaining in the interview.

 

[Dannyalcatraz]

There is no "accounting shell game" because this is not from their filings: Dario just said, on a podcast, that if their current models were stand-alone products, which they are not, those products would be profitable.

You can be done if you wish.

That said- since you responded to an earlier, unedited version of my post- I will reassert that there is an accounting shell game going on.

In the post-edit version*, I included inference as one of 3 major costs Anthropic is incurring, alongside R&D and training. Inference is the portion of AI development in which trained AI is asked to provide responses to new, unseen data.

Inference is also noted specifically as one of the reasons Anthropic is in the red. You cannot have an accurate P&L statement for a product if you’re not properly associating it with one of its major costs.

Sources state that while Anthropic’s revenues DO cover inference costs, they have not covered their products’ R&D or training costs. IOW, they’re ignoring 2 of the 3 major costs of their AI to call it “profitable”.

That’s improper accounting by definition in GAAP,** especially with SaaS (Software as a Service). And Anthropic’s AI is definitely SaaS. If you don’t include a product’s R&D costs, you can’t calculate ROI, etc.

IOW, his claim that their products are profitable is a distortion that ignores the general standards of accounting and the industry standard for accounting in Anthropic’s field.

Or, to put it in terms you said you initially expected from me, yes, he’s lying.


* I had to look up the proper term, and was posting while in the process of driving family around town.

** Generally Approved Accounting Principles

 

[1QD]

What I'm suggesting is that your game sounds like a reaction to D&D - 'like D&D but with more X and less Y'. That's what those games were, back in the 70s and 80s. You might want to look at those games to see if you are inadvertently retreading the same ground they did, or alternatively to see whether their innovations and developments are what you are looking for.

None of your posts have ever been removed. Do NOT challenge moderation. Especially by making things up. That’s the fastest way to be asked to leave. I hope this is clear. If you are not clear about the rules here, feel free to re-read them. And never, ever, lie about having posts removed again.

I am making no attempt to make things up, nor am I lying. My post was moved as we can see here. I am not trying to be combative, but you are falsely accusing me of things that actually happened. It seems you are indeed hell bent on misrepresenting me and I will accept in truth what your motives are. Regardless of the out come it is obvious to me at this time what your motives are. I wish you luck on your endeavours, and screen shot for the win. I refuse to be bullied by someone with ill intended motives.

 

[1QD]

What I'm suggesting is that your game sounds like a reaction to D&D - 'like D&D but with more X and less Y'. That's what those games were, back in the 70s and 80s. You might want to look at those games to see if you are inadvertently retreading the same ground they did, or alternatively to see whether their innovations and developments are what you are looking for.

They seem a bit base in comparison. A great deal of thought went into my system. I see some effort to stream line things and 2 common denominators, other than that they are worlds apart

 

[1QD]

Since this will most likely be my last post. I wish you all the best. A picture is worth a 1000 words. Know at least that I acted in all good faith.