DeepSeek No Travel

| 0 Comments

The Information is reporting that some DeepSeek employees are no longer allowed to travel after the company’s rise to fame. “Management asked some staff to hand in their passports, the three people said.”

This is a part of why I’m bearish on China long-term under the CCP: you can’t enough capital and top talent when you operate this way. If you were the best and brightest picking a university, would you (a) stay in China, where success means restrictions on freedom, or (b) go abroad where you are essentially unlimited?

OpenAI Threatened

Related: OpenAI asked the US government ban DeepSeek because the models are “PRC-produced.” While the diagnosis is probably not far off the mark, the prescription is farsical. There is no legitimate reason to ban the models at present.

…although my own research has shown censorship in models and other research has shown it is trivially easy to finetune existing models so they produce malcicious output.

Don’t Forget Your I’s

Don’t Forget Your I’s

| 0 Comments

The ultimate spam marker: the inability to use personal pronouns in communication. I got this one today:

Noticed you spoke with us last year about leveraging ai and automation on your helpdesk. Is that still a thought?

Shooting you this quick message to connect and see if it’s worth reconvening here in Q1 to chat updates. Let me know!

In an effort to not sound like it is all about the salesperson — which it is — they avoid adding a subject at the beginning of the sentenance, “I”.

Want to sound professional? Be willing to give the ask with a short value prop. I’ll probably still say no, but at least you showed confidence.

OpenAI: Up & To The Right

OpenAI: Up & To The Right

| 0 Comments

This gem came in my inbox this morning from The Information:

Still, OpenAI’s business dwarfs competitors like Cohere and Mistral. OpenAI is expecting to more than triple revenue this year, from $3.7 billion to $12.7 billion, we scooped Friday. OpenAI told investors that it forecasts most of its revenue for the next two years to come from ChatGPT—showing how lucrative the market for AI applications can be.

OpenAI is probably still burning cash, but this type of revenue increase shows the demand for AI. Not only that, it is a proxy for the value people see in it. You don’t plop down $240 – $2,400 annually just for the “fun” of talking to an LLM.

OpenAI is also seeing this huge increase from ChatGPT, the online application version of their LLMs, which is different the paying for API access to their LLMs.

Upwork Content Moderation Is Still Awful

Upwork Content Moderation Is Still Awful

| 0 Comments

Yes, I’m here to complain again in the hope that someday Google will rank these posts high enough someone at Upwork notices.

Today I am hiring for copy-editing and fact checking an eBook. I’ve done extensive work on this already, but want another set of eyes. My job description included links to the documents in question, and I wanted to know if the applicants had reviewed them.

So I sent this message:

Policy alert notification indicating a message was removed for including contact information, violating terms of service. Below, a conversation with Jonathan Addington asking about document review, total hours estimate, and meeting a 48-hour deadline.

And as you can see, it was blocked. Changing the top line to, “Did you look at the documents in the job description?” got around the block.

Their chat support is clearly LLM-only, portion of that:

[Upwork] If your messages are being blocked and you believe they do not contain contact information, there might be another reason triggering the block. Unfortunately, we cannot disclose the exact part of your message that was flagged. However, if you feel there should be an exception for your case, we can look into it further. Would you like to proceed with exploring an exception?

[Me]: I didn’t provide contact details at all
[Upwork]: Thank you for clarifying. If you believe your message was blocked without including contact details, and none of the exception reasons apply, it might be due to another aspect of the message. Unfortunately, we can’t specify what triggered the block. If you have any more details or specific concerns, feel free to share, and I’ll do my best to assist you further.

Moravec’s Paradox

Moravec’s Paradox

| 0 Comments

(I realized only when I was almost finished writing this post that all I had really done was to restate Moravec’s paradox: “it is comparatively easy to make computers exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility.” Or as Steven Pinker put it, “the main lesson of thirty-five years of AI research is that the hard problems are easy and the easy problems are hard”.)

Timothy B. Lee quoting Steve Newman.

Most Read