Something that bothers me with hn is that despite the amazing shit we see with AI, a big chunk of it has somehow decided that the corpo bloodsuckers are also amazing and should be protected at all costs. I wish we could have more nuanced conversations, considering the majority here think of themselves as some sort of high iq individuals.
A great number of posters here are pseudo-intellectuals who think that some expertise in developing CRUD web apps, which they consider to be a form of “engineering”, translates into expertise in all fields.
It made me sad how it destroyed so much cohesion and eliminated people's jobs. I don't find that amazing. It is killing people literally now. Not everyone will be able to find a new job.
Just look at this guy - he retired mostly due to AI taken over ruby, as one example of many more:
And that's about a month ago only, aka quite recently.
AI changed everything. Mostly for the bad. In a few use cases one can say
that AI led to interesting changes, e. g. discovering bugs in projects (assuming
here these are real discovers about real bugs, but quite many were real bugs indeed).
This led to the death of all bug bounty programs. It makes no sense to pay humans when computers can do the same in a better way.
Many here are overconfident. They believe that expertise in one area means expertise in all areas.
Just browse any thread about renewable energy where the tech bro crowd believes that every challenge is a “solved problem”.
Were that the case, private industry would be building data centres powered by nothing but wind and solar energy. Yet, the preferred power solutions still appear to be the power grid or skirting emissions regulations via gas turbines.
> Google is not "cutting off Search from AI-free schools". They're refusing to supply AI-free Search. It takes less theorizing to understand if you phrase it this way.
can someone smarter than me explain in simpler words, how is that literally not the same thing?
Search is already chock full of AI, removing it would be tons of work that Google doesn't want to do (since P0 is in fact cram even more LLMs into everything).
And this is were the misconception appears. LLM's are a type of AI, but AI != LLM. The top part of the results, also called the AI summary. Is _not_ the only thing that is AI about search. Embeddings, a commonly used element of search, or ranking algorithms, use AI as well. More specifically, the techniques are known as machine learning, but it is AI.
If you truly want to have no AI at all, then you are back to boolean retrieval systems. That basically boils down to: Count the amount of times the terms of your query appear in the document and return the top x results (which is impossible to do at Google's scale).
> That basically boils down to: Count the amount of times the terms of your query appear in the document and return the top x results (which is impossible to do at Google's scale).
It is very possible to do at google's scale. Jeff Dean and co built Map Reduce
to help implement algorithms of this type. For example, https://en.wikipedia.org/wiki/Tf%E2%80%93idf is entirely based on counts. Combined with a sparse encoding and distributed lookups, this is how ALL retrieval worked for decades. In fact, any large scale retrieval system today uses a combination of sparse searches followed by a ranking or scoring with dense methods. You can roughly call sparse methods count-based and dense methods neural network based.
> And this is were the misconception appears. LLM's are a type of AI, but AI != LLM. The top part of the results, also called the AI summary. Is _not_ the only thing that is AI about search.
Why post this? The AI summary is the part that schools are talking about. Why post about other irrelevant AI in Google that's not the subject of the discussion?
Are you sure about that? Lots and lots of people would say “anything AI is bad”, even if that’s an ignorant statement. If Google just removed the AI box for schools there would be an outrage the next week when the “story” broke that Google uses AI in other places besides that one box.
This is probably Google's excuse. They are also using machine learning in the search part of search (and have been since before 2020) and choose to maliciously comply by disabling both of them, even though the schools just want the big LLM box gone.
If Google's switch promises to disable the use of all "AI" techniques down to the implementation level, that would imply it also disables spam filtering in gmail. I haven't seen that reported and it isn't in the original announcement post, so it probably isn't how they're thinking.
Though note that before AIOs Google did also have LLM-powered text box results ("web answers") and nobody seemed to have a big problem with those... It seems a bit hysterical to demand them removed when they are just a better version of a long standing feature.
You’re right that there are other kinds of AI frequently used in search, but why do you claim having an inverted index is impossible at Google’s scale? How do you think exact term search works?
If search is that tightly integrated with the LLM capabilities isn’t that an architecture failure? LLM distends rely heavily on information retrieval for grounding their responses. Why would it be so difficult to let humans continue to directly use the information retrieval systems (classic Google) and keep the LLM systems optional and loosely coupled?
Language processing is literally what search does. They're probably using LLMs for it and interpreting "don't use LLMs" as "either make a new search or disable it" when the intention is clearly "don't display the huge trash pile"
The main thing they don't want to disable is the AI summary at the top of the search, which eeeeh, can be seen as AI but won't let kids enter an AI mode /chat like some news stations are claiming. (Based off reporting I watched a week ago)
Google search AI summary accessed from my region does include a text field bellow it. Using it moves you to conversation page, quite similar to Gemini although a bit more minimalistic.
Imagine you have an otherwise AI positive school that wants AI-less Google. Google wouldn't give it to them, but this fact is not communicated by the less precise headline.
Google are avoiding an à la carte offering here. They are saying you either get search which now includes AI, or you can disable the entire feature. The obvious thing that people want is the original search without AI included.
They are not giving reasons why they can't unbundle these options. It could be technical but seems more likely that they want to appear to comply and shift the focus. Search being disabled creates pressure on those operating the tool. The local narrative becomes that the admin is the one that is limiting use. Eventually people lose sight of the real reasons and Google are on the sidelines watching users fight it out with admins.
This is a classic product management approach to dealing with objections. Remember that the PMs are getting measured on usage. If no one uses their feature then they are in trouble, so they make it hard to disable their features. A wonderful incentive structure, right?
Lots of companies build toggles that bundle up features as a result. You may want to disable this one thing, but toggling that off disables all these other things you want. The admin has pressure from the business or users. The vendor comes in to support the business or users, and then the IT team is overruled. Alternatively, the bundle is packed with items of real value plus mitigations are released elsewhere in the product (think: eDiscovery/logging/compliance) that cause IT to enable the whole bundle.
Google search has involved various forms of AI for a decade or more...
All those ranking algorithms that decide which link comes at the top? They use AI.
Sure - it doesn't look like a chatbot, but it's still AI under the covers and they couldn't in good conscious even give you search from 2008 and claim there was no AI involved.
I find these arguments absurd. You know full well what the schools mean. And that it is not the same thing as objection to 2008 technology. Which was not even called AI in 2008 but technically falls into onw of definitions.
The thing the schools want isn't well founded. Presumably they want their students not to be able to cheat, but a perfect search algorithm also allows them to cheat. As does, e.g. Wolfram Alpha-esque deterministic natural language systems.
It really isn't. For example, nobody complained about the previous LLM-based summaries at the top of Google ("web answers"). What is it about AI overviews that is so much worse than them, other than culture war aversion to the word "AI"?
Plenty of people complained about those. The complaints were numerous and loud enough that Google added a "was this helpful?" link and a feedback form for helping them fix the terrible accuracy. They ended up removing it for vast subject areas.
The amount of apologizing by people here on Google leadership's behalf not desiring to implement a feature flag for this population is confusing. This is classic bullying by Google, against kids.
There's nothing confusing about it. Remember that half or so of the HNers reading this thread have worked for Google in some form, or have a partner who has.
That's a lot of emotional appeal -- people who believe there's a reasonable explanation are "apologizing" (which is bizarre, because they obviously are not), Google is "bullying" and of course the old think of the children.
There has to be a better way to disagree with people?
For a group that creates software we all know that feature-flags are trivial and already something Google has in place to serve its varying markets. It is unreasonable to expect that one of the most advanced companies will use this to serve the legal needs of one populace, but not use the already built feature to serve children in a more ethical way.
Sure, I could argue we should all think on how to better help a surveillance and advertising company get a better and earlier grip on children so that they are hooked on AI summaries instead of critically thinking and evaluating sources for themselves. But this is absolutely bullying, it's literally the tactic at use.
>
For a group that creates software we all know that feature-flags are trivial and already something Google has in place to serve its varying markets.
Most software developers also know that feature flags easily leads to combinatorial explosion and spaghetti code. From this alone, I can understand well why Google is not eager to implement such a feature flag.
> Most software developers also know that feature flags easily leads to combinatorial explosion and spaghetti code. From this alone, I can understand well why Google is not eager to implement such a feature flag.
What in the world are these arguments?? It's a single flag.These are the most bent over backwards absurd attempted defenses I've seen.
It would be very expensive to create a custom version of search for these schools. It's not simply a matter of flipping a switch off. They decided that the cost of accommodating the schools wasn't worth the effort.
What we are talking about is the first AI summary page. Simple flag, include it or not. And it's not something that belongs in an educational environment unless you're teaching about the risks of AI. Kids will not have the experience to judge whether it's reasonable to trust it or not.
it's interesting the further the planets are the more off level their orbit is, is that because most of the planet creation chaos was happening near the sun?
It requires less delta-V to create a large change in inclination when the object is far from the Sun, because the orbital speed in a more or less circular orbit varies as (orbital radius)^(-1/2).
Yes, imagine a highly elliptical orbit and you've just reached the furthest point , like the roller coaster at the top of the hill, almost stationary. At that point you could pretty much boost the object in any direction you wanted, and it would then assume that orbit.
Yes. In that situation, where the perihelion is at the minimum tolerable distance from the Sun, the speed at apohelion is proportional to 1/(distance to sun) (this follows from conservation of angular momentum).
I have sent to ~22-30 million requests to OpenAI's completions api since early 2023 asking to classify pieces of text. I'm pretty sure many have sent even more.
I think they just didn't have the incentive to offer a cheaper version that does the same thing.
reply