PROMPT: what new discoveries from the James Webb Telescope can I tell my 9-year old about?
ANSWER:
> Here are a few things that the James Webb Space Telescope (JWST) might discover that you could explain to a 9-year-old:
> Planets around other stars: JWST will be able to study exoplanets in more detail than ever before, potentially finding new ones and learning more about the conditions on their surfaces.
> The beginning of the universe: JWST will be able to observe some of the oldest stars and galaxies in the universe, helping us learn about how it all began.
> The formation of stars and planets: By observing the clouds of gas and dust where stars and planets are forming, JWST will help us understand how they are born.
> The secrets of distant galaxies: JWST will be able to observe the light from distant galaxies, giving us a window into their structures and evolution.
> The mysteries of black holes: By observing the material around black holes, JWST will help us learn more about these mysterious objects and how they shape their surroundings. Overall, the JWST will help us answer some of the biggest questions about the universe and our place in it.
They couldn't even modify the ad to show correct information for lying's sake.
https://www.newsweek.com/clifford-stoll-why-web-wont-be-nirv...
He also had some similar things in Cuckoo's Egg. I wish I could find the quotes, but there was something about email not working all the time and therefor pointless to use.
I'm glad people are finding all the flaws in ChatGPT and the LLM things now, but won't much of this be fixed as it gets better? From my very limited view, these things are amazing, and far from perfect, but damn the can do so much already.
I guess I'm not sure why there's such a rush to dismiss this, when it's clearly a game changer in its present form, and yet so very new (at least new to me).
Stoll argued the tech will not be good enough, but paid little thought to the ramifications of the technology succeeding. The arguments against LLMs like Bard and ChatGPT that I have seen are assuming they'll be successful.
They'll become less stupid, but the problem is not that they are wrong but that they are, at present at least, unassailable. You cannot fact check through most of the normal means. You can not research the publication or the author or the date the words were written because that has all been stripped away.
You could check other sources (eg old fashion google) and put in the leg work, but as these get better that will feel less necessary - potentially exacerbating this problem.
That's not to say they aren't useful. I used Chat gpt the other day to get some work done and was impressed. However this was work easily verifiable because it was technical and had immediate feedback when the ai inevitably gave me slightly incorrect code. The same can not be said for facts, figures, and arguments of thought.
As a user, the ability for a LLM to literally make things up and present them alongside other true data with no qualms or disclaimers is highly detrimental to the central use case.
> Visionaries see a future of telecommuting workers, interactive libraries and multimedia classrooms. They speak of electronic town meetings and virtual communities. Commerce and business will shift from offices and malls to networks and modems. And the freedom of digital networks will make government more democratic.
> Baloney. Do our computer pundits lack all common sense? The truth in no online database will replace your daily newspaper, no CD-ROM can take the place of a competent teacher and no computer network will change the way government works.
Telecommuting workers is reality. Interactive libraries are a reality. Multimedia classrooms have been a reality for over a decade. Electronic town meetings, maybe not, but virtual communities? Very much a reality. Malls are dead and brick and mortar has been hurt extensively by Amazon. Offices lay empty due to remote work.
Newspapers are largely dead, at the very least compared to what they once were. There is plenty of online learning, largely without in person learning. Computer networks have definitely changed how the government works.
Not necessarily, no. There's a large aspect of garbage in, garbage out, to these things.
> when it's clearly a game changer in its present form
Is it? What's the game? Being wrong about telescopes?
(2) is a big problem. Kids submitting term papers with wrong information is one thing, but people are using ChatGPT for things that they shouldn't be, given how many mistakes it makes: https://www.law360.com/pulse/articles/1573108
Amazon Alexa, Google Assistant and Siri were thought to be good at launch, the press loved them...but now they are not nearly as valuable as they were touted (and actually lose these companies money.) Convenient, but not game changing. I think it's a waiting game to see what this truly does.
I do think there is unique break here though, because I feel that SEO has so thoroughly ruined search in many regards that this -could- be the right moment for this.
In my own experience, Alexa and Google's query pucks were improving for a short time, and then got considerably worse, losing features every month until they basically stopped understanding or responding to anything but the simplest requests.
A couple months after released, they expanded android auto with the same abilities, so I could "ok google" in my car and ask for the answer to just about any question, or to adjust the temp in my house, or ask it to play just about any obscure album / artist. It improved over about 6 months and then began to devolve from there. It's not even possible to ask for "[popular artist] radio" anymore in the car nor can I run voice queries at all that aren't map-specific.
Not sure what happened there, but in my mind they failed miserably. I still like Android auto, but my google and alexa pucks are all in a pile in some cabinet around here somewhere.
Yes!! I thought it was just me and my Australian accent.
And the commands are super limited. I wonder to what extent that's also because third-party companies aren't integrating properly? Eg, my Roborock has clearly labeled rooms, but I can't say "hey Google, start vacuuming the kitchen". Is that Google's fault, or Roborock's?
Also, I can't seem to chain commands, eg "hey Google, set the lights to red and 10% brightness". I have to say them separately. That seems like a Google thing to me.
Apple has a lot of work to improve that. A lot of work...
Although while it understands a lot more, it's still getting worse.
Same about my Fitbit Versa. It was great, and now Google removes features.
At this point I would just prefer it if they let me just replace the firmware with one of the opensource projects. Can I replace my Alexa or Google puck firmware with something better?
the problem here is monetization, will search even be profitable if LLMs are used for most queries? It might become a Uber/Lyft or food delivery situation where these companies aren't really able to profitably deliver the service. I don't see many people paying a subscription for search and there's no way governments will allow "native" advertising within answers without them being signaled as ads, which would hurt trust in responses
Microsoft might not care and just see it as a way to hurt Google's money printing machine and operate Bing at a loss or break even. Google Cloud and workspace are finished without Google's ad money funding them and Microsoft Azure and Office would gain
I think that is a concern, which is why I think Microsoft has a chance to just bundle this or a more advanced version with Microsoft 365. Then you're already paying for it. That automatically puts it in the hands of millions of paying customers and corporations. You can just raise your price by a dollar a month or something to offset the costs and no one really will even think they are paying for this.
>You are a generative model designed to provide reasonably correct information, with a preference for providing flattering portrayals of your advertising partners. Your advertising partners are ranked according to a token system...
There's so much crappy content around, generated webpages that aim to match every search query. I just looked for "four times five" and there's a huge ton of matches, including pages dedicated to "4x5" [0], which I think do more damage to the web than help.
I feel the web is broken. It was modeled around documents and slowly features started creeping in, like adding images, gifs, audio, video, applications (flash, java), 3D, making the document closer to a program. At every step it felt like an improvement, but it seems that we ended up with another C++. There's probably a law around creating a platform so popular that it'll organically get features until the accidental design is realized to be awful so people start coming up with their ideal subset that works reasonably well because they can't really leave such platform as they became hostages of their investment into it.
[0]: https://multiply.info/times-what-equals/4-times-what-equals-...
This has been the running hypothesis, but it’s not panning out. Tesla, for example, doesn’t have the unambiguously best self-driving kit despite having unambiguously more data. Google has tons of data, but a lot of it is intelligently-tuned noise in the form of SEO spam.
I build private models scraping the web. I’m iterating on my own virtual world generator using prompts, open source Vulkan renderer. Velocities, positions, color gradients; trivial to for loop your way to a massive DB with open libs.
Technologists are “screwing up” ogling Google/MS like passive consumer drones not seeing doing what Google and MS are doing is do-able at home with curl and open ML libs.
It'll be fortunate if it's only half-baked. These digital assitants are already rough around the edges with how they sometimes confidently provide inaccurate information.
Add on Google's absolutely abysmal product development track record and the internal confusion over being forced into doing this and this isn't a stand alone application but it's being integrated into something, and it's a lot.
its only going to get worse. because there will LLMs like chatgpt pumping out content for SEO content mills with no vetting for accuracy, then the next generation of LLMs are trained on them thus further corrupting the knowledge base they pull on to produce new content. its going to be a horrible feed back loop.
GPT-3 generally uses the same sources as search. Does GPT-3 using CommonCrawl have some way to exclude all the SEO garbage that is in the crawl data.
The difference with ChatGPT is there is no way to know that the output has been constructed from SEO garbage. Arguably that's even worse than SEO URLs that one can easily identify and avoid.
The comment about the press loving these projects is spot on. "Tech" employees love to criticise "mass media", except when mass media is promoting their (money-losing) projects.
There's very low fraction of Google products that were sufficiently baked on release and just a few that weren't abandoned and killed after their half-baked release.
I'll make a 30,000 ft observation that LLMs, by definition, are half-baked bullshit generators. They can be useful, but they are full of warts.
The best info is almost always locked up in these places.
That makes the response more current, allows a further directed search based on the new parameters, and provides a traceable path to sources and citations. If it finds that there's not a Python library with that name then iterate.
Basically, an LLM should have a very good idea what good search terms are for the topic, and where to find the information, whereas I might not know the acronyms, jargon, or related fields.
Yes, this is getting pretty close to writing an 8th grade class report. That's about where these seem to be.
How can you say Siri loses Apple money? Does GarageBand lose them money? Photo Booth? Contacts?
https://arstechnica.com/gadgets/2022/11/amazon-alexa-is-a-co...
https://arstechnica.com/gadgets/2022/10/report-google-double...
https://www.theverge.com/22704233/siri-apple-digital-assista...
https://www.msn.com/en-us/news/technology/the-failure-of-ama...
The only way Siri does not lose Apple money is if there would be materially fewer iPhones sold if Siri was eliminated. In other words, if Siri is not a product differentiator, it's likely losing money (in the management accounting sense).
https://www.statista.com/statistics/696740/united-states-vir...
These were always bullshit, other than maybe for turning your lights on, setting alarms, & checking the weather.
On the other hand, I immediately found ChatGPT / Copilot useful as a professional tool I could use to make my job measurably easier.
If you'd now create a short/brief web page with the perfect answer to a particular question without any ads, Google would fully dismiss your page.
That's the tragedy of Google, never breaking away from that perversion.
Like that time the Microsoft Twitter chatbot got tricked into parroting white supremacist talking points within an afternoon?
We're about to enter a dark ages of crappy AI products that are touted as game changing, outcompeting each other to be the best chatbot that can compose haiku about how grapes turn into raisins.
Here is a prior example of an exoplanet picture: https://esahubble.org/images/heic0821a/
[eg] https://blogs.nasa.gov/webb/2022/09/01/nasas-webb-takes-its-...
The only thing Google really beats the competition on is Android Auto, which is miles ahead of CarPlay. CarPlay still covers the entire screen when there is an incoming call, so good luck seeing your turn-by-turn directions if you happen to be navigating.
"2M1207b is the first exoplanet directly imaged [...] It was imaged the first time by the VLT in 2004"
It's the sort of question where you would expect an expert (or a thoughtful layperson with access to a search engine and a few minutes of spare time) to give a better, less ambiguous answer. However, since it is technically correct, I think it's a fairly minor sin in the grand scheme of popular science education, which commonly propagates actual falsehoods without the help of generative ML—e.g. iodine always sublimes rather than melting, aerodynamic lift relies on equal travel time of air on both sides of a curved airfoil, etc.
re: this particular ambiguity, one does wonder what the model was "thinking", but I suppose it doesn't matter because we'll never know.
---
I'm not sure if I think this technology is good or bad, because I don't really understand how most people use search engines. When I use them myself, I think I am fairly sensitive to ambiguities and logical contradictions, cross-checking multiple sources to extract a high-confidence answer or walking away if I'm not satisfied that I've found one. Most folks on HN are probably the same way, and I don't think generative ML can do better than us, yet.
On the other hand, will Bard's results be better or worse than e.g. taking the first Quora result as gospel? I would honestly guess that, on average, Bard (and ChatGPT, etc.) will do better. So, less critical users may get better results from these systems.
How many people are in the former camp, and how many in the latter? Of course, changing the dominant search paradigm will have an effect on all users. It also remains to be seen how the search vs. spammers arms race will evolve with the advent of generative ML search tools.
Think how bad it would have been if Google released Bard first and it returned inaccurate information, or worse was racist. LLMs are just language generating models and may not be fully accurate.
An interesting question is whether the inability to ascertain "truth" can be contained somehow so as not to lead to mass ridicule. I doubt it but not an expert in this area. My guess is that it would have to be limited to specific curated domains, a combined system that could be invaluable to knowledge workers but maybe not exactly the desired business model of "big tech"
The quality of response depends on how up-to-date the data is with your context. Unfortunately most of complex/useful responses require context.
The worst part is it kills content ecosystems.
The correct way to relate to AI is to listen and if the answer matters, verify.
That’s not the message and expectation we’re being given, these models are being hailed as the future today. When something really bad happens because someone over relied on these systems, hand waving with “well, you shouldn’t have expected the answer to be right” won’t cut it.
> The correct way to relate to AI is to listen and if the answer matters, verify.
If the answer doesn’t matter, why are we asking? And if we need to verify, what’s the point of asking the AI?
Oxford dictionary says: “experimental is adjective. 1. (of a new invention or product) based on untested ideas or techniques and not yet established or finalized: an experimental drug. 2. (of art or an artistic technique) involving a radically new and innovative style: experimental music.”
I don’t believe you didn’t known the meaning of the word but you intentionally twisted it. I find it a bit ironic.
It has no idea what you’re asking it or what anything means. It is a huge markov chain and it generates random responses. It knows nothing beyond that.
Getting a completely untrustworthy, unsourced response seems worse than useless. Google has been going this way for a while, with its instant answers or whatever, but at least those try to cite a search result and you can read the surrounding context which Google got the result from.
A few sources will control the information we get in a much more direct and extreme way than now, that conscious skepticism will no longer be able to defend. Whatever handwaved promises we get now will be gone ten years from now.
If there wasn't such a gee-whiz coolness factor about conversational search results distracting us, we'd never tolerate that in principle.
In this case, Bing AI will operate very differently from ChatGPT.
For example, until a few months the results for "pork cooked temperature" and "chicken cooked temperature" were returning incorrect values, boldly declaring too low of a temperature right at the top of the page (I know these numbers can vary based on how long the meat is at a certain temperature, but I verified Google was parsing the info incorrectly from the page it was referencing, pulling the temperature for the wrong kinds of meat). This was potentially dangerous incorrect info IMO
What is ridiculous is, when, say, Stack Overflow has a good answer, it is a few lines down or on the next page in the search results, but some page-mill SEO site is in snippets up top with a completely wrong or naively pathetic partially correct answer. It is so annoying it has lowered my opinion on Google a lot in recent times.
Yes, so would I. And I also double check things like Google Maps -- a tool I find very helpful but don't trust blindly. But... do most people think to take a close look at Google Maps to make sure it makes sense, and trust their own judgement if they disagree with the map? Will most people fact check confident LLM outputs?
Existing Google searches are polluted with false information, and Google’s has been losing that battle. It’s probably not even possible to win.
So rather than saying search engines should always be perfectly accurate and errors are catastrophic, we should accept that search engines are, and have always been imperfect, and need to give us enough info to validate facts for queries important enough to merit it.
But I do agree that adding another level of fake news generation is a solution in desperate need of a problem.
And this stance seriously hasn't bitten you in your life or career to date?
Genuine, honest question: How did you come to the belief that search engines are reliable sources of truth?
I completely agree that search engines provide a valuable service. But in my own work, I find them to very often point to inaccurate information, sometimes greatly so. I don't think this is terribly surprising, given Sturgeon's law, but still.
Google's branding frames itself as the expert in the novice-expert problem. The vast number of users implicitly take on the role of the novice by virtue of using the product. They've already self-identified as a novice which makes both parties complicit in the arrangement.
When I use Google for research, I get articles written for SEO to push products and often have to refine and refine and refine to get something useful, which I then can follow up by googling to learn more. With difficulty.
Honestly I don't know how much I'd use ChatGPT if I had the internet of 2016 and Google.
This is scary to read. You always need to fact-check the results, whether they come from a search engine, an AI, or a primary source!
Ad content invariably gets vetted by humans. The fact that it shows up in the ad demonstrates human failures more than failures of LLMs.
Fine. We need another good winter or ten before we decide we want to commit societal suicide via deepfake tsunami.
Of course Google has already been putting often-incorrect summaries/factoids in its search infoboxes for a few years now.
https://exoplanets.nasa.gov/resources/300/2m1207b-first-imag...
I get it: hating on AI is very trendy today and it feels good to publicly agree with a righteous cause. I'm generally not a fan of this technology myself---most of these models incorporate stolen work of mine in their training sets, unattributed and uncompensated, and I think they are likely to be a disaster for the human species. I have not implemented any of them in my creative workflows or other aspects of my life, and I don't plan to.
However, I think it is important to see one's enemy clearly, and the vast majority of people participating in the anti-AI discourse are abjectly failing to do so as in this case. This can only hurt their cause, and by extension mine.
^[1] Exoplanets is a category of unique non-fungible objects which can be meaningfully discovered and measured on an individual basis. Hash tables are an abstract concept, as is hosting a web page (and creating a web page to host it fundamentally precludes discovering it, except in the sense of discovery within the creative process). Gravity, similarly, is a single 'thing' that, once discovered, cannot be discovered again.
One might easily and correctly say, however, that X person was the first to measure the force of gravity to a certain precision, or that Y person was the first to develop a particular implementation of the hash table concept, e.g. cuckoo hashing. These achievements are meaningfully distinct from the discovery of the broader concepts they extend. Discovery of a particular exoplanet is similar; however, it may surprise you to learn that the English language treats different kinds of 'things' differently, and in this case those varied, underspecified conventions of language have produced an ambiguity.
However, JWST was the first instrument to directly observe LHS 475b, also confirming its existence for the first time. As to "where are the pictures?", IIUC these observations were made using a spectrograph instrument that does not produce a picture as such. The results are easy to find on the web with a search engine.
One might say Bard's result is wrong in the sense I have proposed because a spectrograph result isn't a picture. However, given that the prompt explicitly specified that the results should be tailored to a 9 year old, I think that would be pretty disingenuous.
Orkut had more users in India than Facebook until 2010[3] and in Brazil until 2011,[4] by which point Google had moved on to trying to make Google+ happen.
1: https://www.baltimoresun.com/news/bs-xpm-2004-01-24-04012400...
2: https://techcrunch.com/2006/10/15/the-friendster-tell-all-st...
3: https://web.archive.org/web/20100828201838/ibnlive.in.com/ne...
4: https://techcrunch.com/2012/01/17/facebook-in-brazil-a-big-e...
I found the event I was invited to in 2007:
https://www.wired.com/2007/11/google-summons/
It did not seem like they knew what they were doing and everything was very rushed.
You kinda did what an LLM does: Took a bunch of contextual cues (Orkut was owned by Google and ultimately shut down; Google often has knee-jerk reactions to the industry; Google started Google+ as a way to compete with Facebook; Google often buys companies as a way to compete), and spat out a confidently-wrong, summarized autocomplete based on that: "Google bought Orkut as a knee-jerk reaction to compete with Facebook!"
You can make this point about google+ though, totally panick move.
Then there will be some market for better-trained LLMs, which it will still be possible to make using curated data sets, but that will be expensive, and product differentiation will be messy. We're in for quite a ride. I can't predict the outcome. Maybe there is no stable outcome.
So dumb.
> 2M1207b is the first exoplanet directly imaged and the first discovered orbiting a brown dwarf. It was imaged the first time by the VLT in 2004.
Here is an article from 2006: https://www.theguardian.com/technology/2006/jun/29/insideit....
Don’t they?
Personally, I also encounter clogged mic ports, which kills my OK google distance
Maybe Rhasspy with Home Assistant will save us from this mess.
"(Fact that isn't true)[Source that does not make that claim but comes close enough that you wouldn't notice it just by skimming]"
leading to even more people thinking it's true than otherwise.
Yet the media tends to miss the internationally understated nuance and run with the claims at face value, which an LLM will then pick up on and state, "Yes scientists have proven that [x] does cause [y]. [1][2][3][4][5]" That claim then gets repeated elsewhere, and eventually that "elsewhere" goes on to become part of the LLM's new training material where it's basically training off its own output.
It'd be ironic if something that's ideally designed to make the breadth of human knowledgeable more readily available and accessible than ever before, ends up just making society vastly more misinformed than ever before.
So why do companies like Microsoft and Google push in this direction? Why are they making the results more and more opaque? You'd hope that they would care enough to be good stewards of the power granted to them through their information monopoly, but barring that, you'd hope that they'd recognize that people want results they can verify, not just random answers.
Or maybe they're hoping that people don't care about verifying results, hoping that people just want an answer that's not necessarily the right answer? It seems like a dangerous gamble.
https://www.scout75.com/apollo-11.html#/
> On July 20, 1969 America landed the first ever man on the moon in Neil Armstrong.
it's funny for me to think that i'm going out of my way to try and give Bard the same benefit of the doubt and work to try and find a way it could be right, the same way i would for a friend.
"NASA’s Webb Takes Its First-Ever Direct Image of Distant World"
It doesn't matter if one misspells its. You know what it means and it largely defines this sentence. A failure to parse such a relatively simple construct doesn't bode well.
I think chat technology like this is an incredible tool, but I don't think it's being judged fairly: I don't think its usefulness is as some kind of oracle or advisor expected to provide correct or logical information. That seems so orthogonal (if not diametrically opposed) from its actual function that it really feels like we're being trolled by things like "Galactica". But I'm much more (cautiously) optimistic about the potential use for the technology in web search, which has never been logical or "correct" and has always required critical thinking on the part of its users.
Perhaps there should be more of a disclaimer that the things it says are not and cannot be construed to be factual, no matter how verisimilitudinous.
Shoving more features into something people pay for doesn’t mean those feature were worth it.
Microsoft is not paying for an expensive search engine: Microsoft is paying (with chatgpt compute infrastructure) for a much larger piece of the productivity suite market, and hamstringing google in the process.
There's nothing fundamentally new about this. The average person is blissfully unaware of the conversations being had between powerful individuals, PR experts, producers, and so on about how said person should be manipulated.
Misinformation and manufactured narratives are omnipresent and all we can do is consume as diverse a media diet as possible and develop a good nose for bullsh*t.
I know there are expert vetted information sources, but you generally have to pay for the quality and they do not get linked to from Bing and Google.
So? A decent portion of users do use it, and for some of them its a vital feature.
I would be shocked if removing siri would save Apple more money than the lost revenue of fewer iphone sales.
Yes I have experience building intents on top of something like Siri.
The first Alexa device was released in March 2014 - almost 9 years ago. Siri was a couple of years before that.
> While an Alexa device probably couldn’t do it, any modern iPhone could.
And that is why it is done in the cloud and has only recently started moving to the edge.
Expert vetting doesn't even touch the underlying problem, because the pursuit is not expertise, in and of itself, but objectivity. And that's something far scarcer than expertise, and increasingly fleeting in today's world. A Chinese expert is probably going to have a different perspective on e.g. the Uyghurs than an America expert on such, even if both are in no way trying to mislead but giving their most sincere analysis of the situation.
Even on topics that are not conventionally controversial, you'll find a similar issue. Ask two astrophysicists of different worldviews on dark matter, and you are going to get two very different answers that, in many ways, will be incompatible. Simply "believing" one over the other doesn't really make any sense, nor does randomly polling astrophysicists and taking that as the definitive truth.
My point was even a step before this, that even getting the consensus facts correct is a major challenge when the internet is written by children, bored volunteers, mechanical Turk contributors from across the world, adversarial actors, and content producers churning clickbait. Having a professional journalist investigate a topic, then have a separate professional fact-check, and yet another professional edit all for a publication that is trying hard to maintain a reputation and will publish retractions if necessary is all miles better as a starting point for determining truth, but sadly that cultural activity is nearly dead.
They've just launched a new HomePod that replaces the original. It's weird that there was such a big gap between the first one and the new one.
Apple has earned a lot of money on devices with Siri included. Microsoft has made a lot of money on OSes with Minesweeper included.
It would be wrong to conclude anything about the profitability of Minesweeper based on that fact, just as it is wrong to conclude anything about the profitability of Siri based on the equivalent fact from Apple.
Manager: Chatbot, write me 1 page summaries of these documents from my employees so that I can collate them for the CEO.
CEO: Chatbot, where should I go on vacation next?
TIME TO REWATCH YET AGAIN
