Artificial Intelligence thread
- Thread starter 9dashline
- Start date
before you know it, AI will only do the architecture and companies need code monkeys(humans) to do the implementation and typing...
Azure literally has the weights to opus, why not just inference it themselves instead of paying anthropic the "loyalty tax"? unless Opus is a 5 trillion model and the inference literally does count a mountain of moneybefore you know it, AI will only do the architecture and companies need code monkeys(humans) to do the implementation and typing...
A classic cost benefit dilemma. Consumer spending is down across the world as the middle class struggles to stay financially solvent, and businesses, no matter how fast they're able to move with AI, can only spend what they take in. If the end users of software don't have more to spend, making more features & writing more code is of little use - it's not going to increase sales but will only add to costs. Simply "accelerating" is useless if people aren't buying the products you're "accelerating" towards.
This is also the reason for the large-scale lay offs that have been happening in the IT industry. To pay for more data centers and AI tokens, companies have had to cut back on human staffing, since they're having a hard time just selling more products. But the more jobs you cut, the worse the economy becomes, and the lower consumer spending gets. Ultimately, businesses have to sell their services to someone. The whole market capitalist system revolves around the relationship between supply & demand, and if the demand just isn't there, then no matter how much supply you have, it's useless.
To justify models like Mythos that cost hundreds of dollars for a single query, there has to be a business use case that justifies it. Or I guess the US investors and government can just keep pouring money into the hole, but sooner or later the house of cards collapses. If AGI is reached but it turns out, it is vastly more expensive than human intelligence, it may, in fact, not be viable at all beyond some highly niche use cases.
This is also the reason for the large-scale lay offs that have been happening in the IT industry. To pay for more data centers and AI tokens, companies have had to cut back on human staffing, since they're having a hard time just selling more products. But the more jobs you cut, the worse the economy becomes, and the lower consumer spending gets. Ultimately, businesses have to sell their services to someone. The whole market capitalist system revolves around the relationship between supply & demand, and if the demand just isn't there, then no matter how much supply you have, it's useless.
To justify models like Mythos that cost hundreds of dollars for a single query, there has to be a business use case that justifies it. Or I guess the US investors and government can just keep pouring money into the hole, but sooner or later the house of cards collapses. If AGI is reached but it turns out, it is vastly more expensive than human intelligence, it may, in fact, not be viable at all beyond some highly niche use cases.
this is old news. Been reported several days now. And if you have post something like this, please add some commentary.
Doubao is the most popular Chinese AI app and also gets the most engagement per user.
Probably copyright and hosting agreements, it would be a huge hit to Azure reputation as a compute host to just steal/borrow hosted software services.
Huge capex expansion for ByteDance and it's apparently procuring QCOM AI ASICS also. I presume just put more of them together.
This is just an isolated case; employers can still find other reasons to dismiss employees.