781 karma · joined April 30, 2014
Also see this quote from Ethan Mollick on twitter:
> I have a strong suspicion that “prompt engineering” is not going to be a big deal in the long-term & prompt engineer is not the job of the future
> AI gets easier. You can already see in Midjourney how basic prompts went from complex in v3 to easy in v4. Same with ChatGPT to Bing.
https://twitter.com/emollick/status/1627804798224580608?lang...
But at the end of the day the test was still synthetic!
Placing out-of-context things in a 200k document, needle in a haystack style.
Claude is still very very powerful for extracting data from 200k when it’s real world data and real questions (not adversarial synthetic test).
(posted the image as twitter is not readable in a logged out state, and is annoying to many)
- OpenAI has damaged their brand and lost trust, but may still become a hugely successful company if they build great products
- OpenAI looks stronger now with a more professional board, but has fundamentally transformed into a for-profit focused on commercializing LLMs
- OpenAI still retains impressive talent and technology assets and could pivot into a leading AI provider if managed well
---
Sam Altman's Leadership
- Sam emerged as an irreplaceable CEO with overwhelming employee loyalty, but may have to accept more oversight
- Sam has exceptional leadership abilities but can be manipulative; he will likely retain control but have to keep stakeholders aligned
---
Board Issues
- The board acted incompetently and destructively without clear reasons or communication
- The new board seems more reasonable but may struggle to govern given Sam's power
- There are still opposing factions on ideology and commercialization that will continue battling
---
Employee Motivations
- Employees followed the money trail and Sam to preserve their equity and careers
- Peer pressure and groupthink likely also swayed employees more than principles
- Mission-driven employees may still leave for opportunities at places like Anthropic
---
Safety vs Commercialization
- The safety faction lost this battle but still has influential leaders wanting to constrain the technology
- Rapid commercialization beat out calls for restraint but may hit snags with model issues
---
Microsoft Partnership
- Microsoft strengthened its power despite not appearing involved in the drama
- OpenAI is now clearly beholden to Microsoft's interests rather than an independent entity
- Sam complained that Helen Toner's research paper criticized OpenAI's safety approach and praised Anthropic's, seeing it as dangerous for OpenAI
- He reprimanded her for the paper, saying as a board member she should have brought concerns to him first rather than publishing something critical right when OpenAI was in a tricky spot with the FTC
- He then began looking for ways to remove her from the board in response to the paper
---
Helen Toner's Perspective
- She believes the board's mission is to ensure OpenAI makes AI that benefits humanity, so destroying the company would fulfill that mission
- This suggests she prioritizes that mission over the company itself, seeing humanitarian concerns as more important than OpenAI's success
---
Microsoft Partnership
- The Microsoft partnership concentrated too much power in one company and went against the mission of OpenAI being "open"
- It gave Microsoft full access to OpenAI's core technologies against the safety-focused mission
---
Governance Issues
- The conflict shows the adversarial tensions inherent in OpenAI's structure between nonprofit and for-profit sides
- The board's mandate to act as a check and balance on OpenAI seems to be working as intended in this case
---
Criticisms of Players
- Altman appears reckless in his actions, while Toner seems naive about consequences of destroying OpenAI
- Their behavior calls into question whether anyone should have this kind of power over the development of AI
---
Future of AI Development
- Attempts at alignment and safeguards by companies like OpenAI may be ineffective if other actors are developing AI without such considerations
- Who controls advanced AI is more important than whether the AI is friendly.
- Nationalization of AI projects may occur
it's there to give your perspective
add me to happy customers
If you really want it to be efficient display a static "cached" image of the rating and swap it for the iframe once the image is loaded and displayed. This ensures fast response time and accuracy of the fresh rating at the same time.
The Nytimes website with the exact same look and functionality could be well under 1MB easily.
Most of CSS OM is already implemented in Javascript which can come in handy for this project:
https://github.com/jsdom/cssstyle
(both of which jsdom use internally. cssstyle is a kind of updated version for some parts of CSSOM)
Given 440 is what we’re all used to on an everyday basis, the effects of the OP experiment might be explained by the “novelty” of how 432 sounds to us.
There are tremendous differences between programming technologies. Because the word “language” can distract you here.
What you say in French and German and English will be pretty much the same.
But many programming “languages” practically have very different runtimes, idioms, guarantees, caveats. For example in some all your data is immutable. Or take a look at the concurrency model of Elixir/Erlang/Go “processes” vs Java Threads).
Erlang has a preemptive scheduler for those “processes”. Try doing that in vanilla JS.
It’s not just syntax. Syntax is the surface difference but it doesn’t end there.
This is not to nitpick the posted website which is a great first step but it won’t tell you the whole story.
Another example is closures. Some languages don’t have them at all. Some languages allow modifying closed over variables. Some treat them as immutable constants (but still accessible to read).
Those are huge semantic differences with profound implications to how you approach problems and write code.
When comparing Programming technologies always think in terms of: how does it represent data, how does it move/share data between entities, how does it handle errors, what guarantees does it have.
Finally an inseparable part of any programming technology is the standard library, and nowadays the whole third party ecosystem of packages.
When thinking about what is different about say Python and Erlang I’d argue the least significant of that is syntax.
You won’t build machine learning models in Erlang and you won’t build highly-available, fault-tolerant, billion-user chat applications in Python.
And trust me not because of how the for loop looks.
Yes and if you click “microScript programming” in the menu you are taken to the language itself.
Didn’t find a way to link to that part but it’s just one click away.
I love how you can move a function definition from a class to outside of a class or even into an object “literal” without breaking syntax.
And same with variables, you can restructure it from a global into a class property or as a property of an object literal.
It looks even more intuitive and forgiving than lua and lua was already a joy
What do you think about our product? interested? very interested? or very interested? which one? which one? which one? https://www.youtube.com/watch?v=EZJQDpS7ikc
With the dislike count gone you cannot make a quick judgement about what Youtube recommended to you. Less power to the people, more power to Youtube.
It's a very subtle change. Also a monumental one.
You know what else you can't be 100% certain of? How many dislikes this video got.
It's one thing how the comment disappears. It's a whole other thing how you are asked to "take a screenshot of the disappeared comment to investigate".
If you feel this singles out Google, think twice. Yes, Google definitely wasn't afraid to meddle with its search results for its own gains before. But there are other companies having practices just as shady.
Like "how the Apple Podcasts app went from a 1.8-star score all the way to 4.6 stars in a month without any actual fixes"
https://www.theverge.com/2021/11/19/22791968/apple-podcasts-...
Hiding dislikes, disappearing comments, tricking users into reviewing unrelated services.
There is a broader issue unfolding recently about platforms and transparency.
But the second step is part of Chromium. Chrome shows the same alert for extensions changing the new tab page for the first time. (Which to me makes sense)
Thanks for such a quality app.
Im only using it to fully turn off my built-in monitor while working in clamshell mode and then manually setting the external’s brightness.
So no sync mode (and not even sunset sunrise)
But solved my problems nicely and no more reaching out to my external monitor’s buttons in Narnia
If I’ll keep using it I’ll make sure to caffeinate you ;)
Instructions video: https://www.youtube.com/watch?v=D_01P4-ubdA
If you get weird blue and yellow circles,
=> turn OFF 'Reduce White Point' in Settings / General / Accessibility / Increase Contrast
Note: it works because you don't need to be a developer to load apps to your iPhone via Xcode anymore
Note2: it uses "private" APIs so it won't be accepted into the App Store (just like f.lux).
Is this only for noncritical, reproducible data as S3 reduced redundancy?
Apple loves to feature apps showcasing their latest technologies.
They show such apps as a good example for other developers to follow.
There's criticism of serving something in a bad way, and then there is calling someone an "idiot".
Should we instantly call every single person an idiot for making a mistake?
I wouldn't say that the food was bad here, it was just a bit cold.
> We live in the "social sharing age". People even share what they had for lunch. If anything, people should be encouraged to share LESS or more meaninful stuff only.
This is off topic, but I don't think people should share less. If you are bothered by it, you should read less. Maybe we should have better tools for filtering. But information is everything, something as banal as "I had a bad day" has value. Tracking social sentiment for example.
FBWhichApplicationAreWe_DO_NOT_USE.h
FBProfileSetEventsCalendarSubscriptionStatusMutationOptimisticPayloadFactoryProtocol-Protocol.h
FBStoryCreateInputDataAttachmentsLifeEventExtrasEducationInfoEndDate.h
FBReactionAcornSportsContentSettingsSetShouldNotPushNotificationsMutationCall.h