dpflan

41 ms·

631.
by dpflan 4y ago
Indeed, the more expert the more one can get out of these LLMs. Curation of the content and questions is critical so the better one is at that, the more useful this kind of quasi-expert system is.
632.
by dpflan 4y ago
Curious: Can someone explain what they are excited to use this for? Can someone provide a large scale use-case/scenario?
633.
by dpflan 4y ago
Social Media viral coefficient may have been higher with a quieter announcement and the CEO just showing up leverage social media managers.
634.
by dpflan 4y ago
I agree, prompt engineering is a fuzzy subset of the possible space that could provide the guarantees.
635.
by dpflan 4y ago
Ha, yes, modding it would make it a more accurate 'fraud'-ulator.
636.
by dpflan 4y ago
Isn’t prompt engineering a human way to explore the latent space of the model to create the mathematical interactions we want/deem useful? A human way of traversing the hyperdimensionalty. Given your background, would you say that prom
637.
by dpflan 4y ago
It pretty much seems like it, as this is the infancy of LLM UX.
638.
by dpflan 4y ago
Is there where ChatGPT and GPT-4 and the like can be helpful? I know a few days the discussion of AI tutors was double edged, but if we focus on the later stages of schooling / career path, can AI tutors be useful for advancing the fie
639.
by dpflan 4y ago
I mean it is going to be extreme amygdala highjacking, using whatever is possible to conjure up specialized (i.e. sexualized) messages that resonating to one's ideological and base biological core. An AI Cordyceps-ection. Low informati
640.
by dpflan 4y ago
I can see personalized propaganda without troll farms being a big use case.
641.
by dpflan 4y ago
What is your take then on how this affect your field? And your occupation? Do you think you will incorporate such technology into your day-to-day?
642.
by dpflan 4y ago
How much wiggle room is there with weights? Tweak a single weight, is that copyright infringement or a new weight set?
643.
by dpflan 4y ago
Nice write up, would be nice to have full company names in addition to tickers.
644.
by dpflan 4y ago
More: Do Users Write More Insecure Code with AI Assistants? ( https://arxiv.org/pdf/2211.03622.pdf ) More relevant discussion, from Andrew Ng's The Batch newsletter: Check the Generated Code Generates Overconfident
645.
by dpflan 4y ago
Can we aggregate all the other ai avatar generators for comparison, why is this one climbing up the front page?
646.
by dpflan 4y ago
Thank you, this is awesome. Here is a link: https://www.wur.nl/en/article/Foodvalley-2.htm
647.
by dpflan 4y ago
The Netherlands is mentioned once with a link another article. If its greenhouse technology is so good, were there any viable innovations building upon their success? There seems to be a lot of informed, quality discussion by HNers, are any
648.
by dpflan 4y ago
Thank you for bringing up the hardware perspective for LLM intelligence development.
649.
by dpflan 4y ago
All right, let’s dig deeper, I like your responsiveness. Imagine if the LLM were randomly provided the same starting point as human brain…what happens? Is evolution a differentiable process?
650.
by dpflan 4y ago
Can one argue that the random initialization of a model is akin to this starting point? A human brain has reached a morphogenetic starting point thanks to much evolution, a process full of randomness.
651.
by dpflan 4y ago
Any new technology is going to be used by hucksters, everything is, religion is. It doesn’t matter what it is. What is the problem here, how do we improve society, such that hucksters don’t need to huck? There are absolutely segments of soc
652.
by dpflan 4y ago
It makes me wonder, what is the energy consumption of a what it takes to nurture a human to be an expert in something? Discussions of the intelligence in the LLMs as an emergent property of differentiating the model on troves of quality dat
653.
by dpflan 4y ago
For sure. Troll farms seems like they will be scaled up even more.
654.
by dpflan 4y ago
Thank you.
655.
by dpflan 4y ago
I think there is a bit of a paradox, debugging can seem heavy, but then when you’ve added enough print statements you’ve spent more time and added more things to clean up than if you had just taken the time to debug, well, you should have d
656.
by dpflan 4y ago
Exactly. What is this?
657.
by dpflan 4y ago
Paywalled, but seeing excerpts here, and I’m wondering, how is this an article in the WSJ today? Your passcode can be observed and thieves will try ways to observe it? Is there more to this?
658.
by dpflan 4y ago
https://news.ycombinator.com/item?id=34909855
659.
by dpflan 4y ago
Agree. You're using products from https://www.withings.com/us/en/ ?
660.
by dpflan 4y ago
Along this topic of Apple Watch, does anyone have a "killer"/"can't live without now"/"actually makes the watch more useful" app they use (other than watchOS defaults)?

More ›