You're the man now, dog!
4,819 karma · joined May 23, 2010
(since AI can now answer many questions that might have been topics of conversation; people can use AI to participate; people may be reluctant to participate if AI can data mine everything and link it back to them, etc. similar to Stack Overflow)
there is no such thing as an AI that is not somehow implicitly aligned with the values of its creator, that is completely objective, unbiased in any way. there is no perfect view from nowhere. if you take a perfectly accurate photo, you have still chosen how to compose it and which photo to put in your record.
are you going to decide to 'censor' responses to kids, or about real people who might have libel interests, or abusive deepfake videos of real women?
if you choose not to decide, you still have made a choice.
ofc it's obvious that Musk's 'maximally truth-seeking AI' is bad faith buffoonery, but at some level everyone is going to tilt their AI.
the distinction is between people who are self-aware and go out of their way to tilt it as little as possible, and as mindfully, deliberately, intentionally and methodically as possible and only when they have to, vs. people who lie about it or pretend tilting it is not actually a thing.
contra Feynman, you are always going to fool yourself a little but there is a duty to try to do it as little as possible, and not make a complete fool of yourself.
- enabling local MCP in Desktop like Claude Desktop, not just server-side remote. (I don't think you can run a local server unless you expose it to their IP)
- having an MCP store where you can click on e.g. Figma to connect your account and start talking to it
- letting you easily connect to your own Agents SDK MCP servers deployed in their cloud
ChatGPT MCP support is underwhelming compared to Claude Desktop.
not indicated, and the general idea of dataviz is to communicate clearly. when you have a number, what it represents should be noted, and the units. if I see a number in that context, I assume it's calling out the value displayed.
the x axis is also a bit off, would ideally plot the date of the release and use a proper time axis.
a title is also good to have, maybe a data table.
sorry to be cranky but those who are downvoting , try to be clear, learn some standards, or stay away from publishing charts. you can even ask AI to clean up your code to conform to a standard. Soft skills are important for an engineer. You need to explain the work in clear, persuasive language and dataviz. or you can be, I'm a super-smart engineer, you figure out what I'm trying to say, I don't need to worry about making your eyes bleed. crikey.
https://www.datavizstyleguide.com/
https://www.amazon.com/Better-Data-Visualizations-Scholars-R...
Looks like a chart crime scene
https://www.wired.com/story/ai-safety-institute-new-directiv...
(oh yay, government is keeping us safe from woke AI...eye roll)
I feel like, with proper UX in the cockpit and on the controller console, making it easy to send/acknowledge the clearance, and intrusively demanding immediate acknowledgment for important messages, with the controller able to talk to the pilot if it isn't immediately acknowledged, structured messages would save time, be more accurate, allow automated checks, i.e. be a superior substitute.
UX needs a ton of work and human factors validation, and would take 20 years to implement. But if you were starting from a blank slate it seems like the way to go!
Like, check in with the controller but most messages are sent electronically and acknowledged manually.
I have your clearance, advise when ready to copy, then you write everything down on kneeboard with a pencil and then manually put it in the navigation system, is a little archaic.
certainly speech to text is a useful transition but in the long run the controller could click on an aircraft and issue the next clearance with a keyboard shortcut. then the pilot would get a visual and auditory alert in the cockpit and click to acknowledge.
I would hope someone at NASA or DARPA or somewhere is working on it. And then of course the system can detect conflicts, an aircraft not following the clearance etc.
these models are being commoditized.
There are significant restrictions on it so it's not fully open-source, but maybe it's only a real problem for Google and OpenAI and Microsoft.
Open source has turned into a game of, what's the most commercial value I can retain, while still calling it open-source and benefiting from the trust and marketing value of the 'open source' branding.
I think it should also run well on a 36GB MacBook Pro or probably a 24GB Macbook Air
client.chat.completions.create(..., response_format={"type": "json_object"})
But the nature of LLMs is stochastic, nothing is 100%. The LLM vendors aren't dummies and train hard for this use case. But you still need a prompt that OpenAI can handle, and validating / fixing the output with an output parser, and retrying.
In my experience asking for simple stuff, requesting json_object is reliable.
with LangChain even! eye-roll, you can't really title the post 'every way' and omit possibly the most popular way with a weak dig. I have literally no idea why they would omit it, it's just a thin wrapper over the LLM APIs and has a JSON output parser. Of course people do use LangChain in production, although there is merit to the idea of using it for research, trying different LLMs and patterns where LangChain makes it easy to try different things, and then using the underlying LLM directly in prod which will have a more stable API and fewer hinky layers.
this post is a little frustrating since it doesn't explain things that a dev would want to know, and omits the popular modules. the comment by resiros offers some good additional info.
sometimes the company is worth more dead than alive, the parts are worth more the whole, especially when you can leave someone holding the bag, and the PE company gets paid to make them dead.
in any event the company is worth more to an extremely unscrupulous buyer than as a going concern in public markets.
But Google says that their mission is to organize the world's information and make it universally accessible and useful. A technology that understands and generates human language, with all its idiosyncrasies and connotations, is probably pretty important to that mission. And OpenAI stole a march on Google in commercializing it.
Maybe they justifiably panicked a little because they were starting to miss the boat.
It's a fine line between running around with your head cut off, and entering terminal decline due to too little too late.
If they didn't and just cloned her voice, it's more disregard for creators and artists than I would have thought possible. What were they thinking?
Edit after reading the official story... not sure I believe it, seems disingenuous, at best they chose someone because they really really sounded like Scarlett Johansson, and no one said, it might be a problem.
https://openai.com/index/how-the-voices-for-chatgpt-were-cho...
Interesting story of how some private equity guys would
- buy hospitals
- sell the real estate for more than they paid for the hospital, signing a long-term lease at a high rent
- pay themselves an immediate huge profit. the higher the rent the hospital promised, the bigger the sale/leaseback deal, so the bigger the profit.
- default, hospital goes bankrupt, the community and the dumb patsy who bought the hospital gets left holding the bag.
classic bustout from Goodfellas or The Sopranos, but mobsters get investigated, PE guys don't.
it's the operating system, some people might think it's a tax on everything, some people might think it provides the foundation to produce everything of value.
similarly, Google is the high-order-bit in the information or content economy, the creators get underpaid, the people who do ad optimization get overpaid.
no financial markets -> no IPOs -> no VC -> no Google and Silicon Valley as we know it.
the closer you are to the money and the transactions, and the high-order bit, the better the opportunities to redirect and organize to your advantage, and the more you get paid.
what's the TFLOPS/$ and TFLOPS/W and how does it compare with Nvidia, AMD, TPU?
from quick Googling I feel like Groq has been making these sorts of claims since 2020 and yet people pay a huge premium for Nvidia and Groq doesn't seem to be giving them much of a run for their money.
of course if you run a much smaller model than ChatGPT on similar or more powerful hardware it might run much faster but that doesn't mean it's a breakthrough on most models or use cases where latency isn't the critical metric?
the relational ACID model is overkill for mostly-read data warehouse and verticality helps; streaming is different; graph dbs are different.
Postgres may not be 'all you need', it will take you pretty far though, maybe it's 'all you need most of the time'. 60%+ of the time it works every time.