7 ms·
HTML is the solution to walled-garden lock-in? What? Those walled gardens already use HTML, including some of the semantic elements mentioned (plus ARIA semanti
by p-e-w 3y ago
HTML is the solution to walled-garden lock-in? What? Those walled gardens already use HTML, including some of the semantic elements mentioned (plus ARIA semantic attributes, which are much more sophisticated).
> ChatGPT-like interfaces are likely the future of human data access.
And the whole point of artificial intelligence systems is that they don't require specialized "machine-readable" annotations in order to process input. ChatGPT (and its future offspring) can navigate regular websites the same way humans do. They don't need us to hold their hand. They know when a sequence of paragraphs constitutes a "list", without it having to be explicitly marked as such, etc.
What the author appears to be describing is simply an API mediated through HTML semantic elements. But if you have an API, you don't need a Large Language Model for automatic data access – a good old Python script using Beautiful Soup will do just fine. And it has the added benefit that it runs entirely locally.
- mmahemoff 3y agoValid point. Ironically the main benefit of semantic markup is now an abstraction to help the human developers effect styling and control.
- jraph 3y agoIsn't the main benefit of semantic markup still accessibility?
- mmahemoff 3y agoThat’s one of the main benefits, but if the machine can make sense of the content, it can still present it however is clearest for the user.
- oneeyedpigeon 3y agoBut explicit, human-verified metadata is always going to beat inferred, fuzzily-extracted data, surely?
- mmahemoff 3y agoMaybe in some cases, but that requires time and expertise. It may not be worth it in many cases.
- theteapot 3y ago> ChatGPT (and its future offspring) can navigate regular websites the same way humans do. They don't need us to hold their hand. It can?! It does that?
- jack_pp 3y agoIn its current form, not as far as I know. Even the OpenAI tutorial on this matter scrapes first and uses only normal text. https://github.com/openai/openai-cookbook/blob/main/apps/web-crawl-q-and-a/web-qa.py https://github.com/openai/openai-cookbook/blob/main/apps/web...
- andai 3y agoChatGPT had a Browser Plugin (via the Plus subscription) but last I checked it was removed, (possibly because it cannibalized Bing? Or just the reliability issues).
- kennyloginz 3y agoYou don’t need a good old Python script using Beautiful soup, you can use an LLM with the added benefit of not dealing with that cruft.
- selfhoster11 3y agoAn LLM is much slower and requires more resources at runtime than Beautiful Soup (if you are hosting yourself), and could fail to retrieve data. It's a worse solution unless your HTML parser/scraper is broken by website changes. This is only the case for now, however. I expect this to change in favour of LLMs as time goes on and they become easier to deploy and better at their job.
- wruza 3y agoChatGPT (and its future offspring) can navigate regular websites the same way humans do It seems to be at least a not-yet-true claim, but let’s ignore that for now. It’s interesting if LLMs actually could do this. As I understand it, LLMs are trained on texts and source code among other things. But lacking… let’s name it a reasoning apparatus, can they really look at a DOM tree and tell what it is/does? It’s not a text, and barely can be a well-structured source code that was ever discussed (it’s a result of either bundling or componentization). This is almost on par with “LLMs will look at any .exe and be able to integrate with it immediately”.
- delusional 3y ago> “LLMs will look at any .exe and be able to integrate with it immediately”. The LLM will look at any .exe and determine if it halts.
- KronisLV 3y ago> The LLM will look at any .exe and determine if it halts. At that point you'd need an AGI that can figure out something we can't. Edit: but yeah, sarcasm. Nowadays I can't even tell sometimes.
- wruza 3y agoI believe it was a positive sarcasm in gp.
- kaba0 3y agoI am 100% sure that parent poster was being sarcastic at the hype around LLMs and how many think they can solve everything, even if said thing is impossible to do, e.g. the halting problem. Do note that the halting problem is fundamentally true, no AGI will realize some new way around, unless are mathematics are flawed to the core.
- mythhabit 3y agoThere is some nuances here. While the general halting problem for a general Turing machine is undecidable, and with a fairly easy to understand proof as well, the computers we run today are not a general Turing machine. They are of a weaker class called Linear Bounded Automatons and for the programs they can run, the halting problem is fact decidable, on a theoretical level due to their finite nature. So we will probably never practically solve the halting for LBAs, but the quest for this AGI that should be able to solve this, is not just day dreaming, it's rooted in the theory.
- fauigerzigerk 3y agoAI in general could "see" a website like we see it. But LLMs specialise in text. They will find it much easier to figure out the correct semantic interpretation of tag soup than deriving the same information from the rendered output like we do. Also, the big question is cost. I think semi structured text will forever be far cheaper for an AI to process than a completely unstructured data stream representing visual information.
- abwizz 3y ago> They know when a sequence of paragraphs constitutes a "list", without it having to be explicitly marked as such, etc. that's the thing; they don't know - they guess
- rezonant 3y agoI got the sense that the article was advocating for using good semantic machine readable content over HTTP instead of (or at least in addition to) JS-only div soup so that automated agents like the new wave of LLMs can easily pull out the important details without spinning up a headless browser and rendering the page first. I have the same interest but for the purpose of crawling and upstart search engines. If indexing every page required running the page in a headless browser first, the barrier to entry for new search engines is a lot higher.
- bob1029 3y agoThis seems reductive to me. There's "HTML" and then there's the kind of website where the final DOM isn't known until the user has already been attempting to read it for 10 seconds. There is a substantial difference in the % of browser capabilities that need to be exercised between the extremes of use. Complexity of implementation is what ultimately separates the good from the bad. Any tool can be operated skillfully or poorly. An apprentice with a circular saw and a fully charged battery can do a hell of a lot of damage. A master may elect to use no tool at all and simply bang on the side of the thing (I.e. push back on the business). The latest websites I have built are some of my most compatible ever. I don't use web sockets anymore. I don't depend on JavaScript to have piecemeal conversation with the server. You can actually use ~80% of the product with JavaScript entirely disabled. How are the engineering choices demonstrated here not exactly the solution for walled garden lock-in?
- feitingen 3y agoLots of websites already present a machine friendly site to google-bot and some other spiders. I don't see why they can't offer the same to other bots and only serve the Javascript-heavy pages to humans.
- MrDresden 3y agoThe point is that a lot can be done without the need for Javascript. Also in pages served to humans.
- flagrant_taco 3y agoThe idea of maintaining two entirely different versions of the same side gives me flashbacks to the days when they sir had a separate "m." codebase for mobile. There's a reason we found better solutions there, building and maintaining the same site twice is almost never worth it.
- oneeyedpigeon 3y agoThis is why you wouldn't build "two entirely different versions of the same site", you'd build one version and toggle the <script> tag on and off.
- yellowapple 3y ago> And the whole point of artificial intelligence systems is that they don't require specialized "machine-readable" annotations in order to process input. I don't require a cup to hold my drink. That doesn't make cups useless or undesirable. Point being: if a machine can make sense of the veritable clusterfuck that is the "pile of infinitely-nested divs" status quo, then surely it'd have a much easier time making sense of pages that actually use HTML properly. If you're an AI trying to figure out how far along something is, which is gonna be a more obvious indicator? <div id="reactElement420" class="wangularClass69"><div class="25-long red" visibility="hidden"/><div class="50-long yellow" visibility="hidden"/><div class="75-long green"/></div> <progress max="100" value="69">69%</progress> The first example is hyperbolic, sure, but only slightly.
- fennecfoxy 3y agoCompletely agree with you. The magic of LLMs is being able to chuck it an unstructured mess of data and have it parse and interpret this data in some semblance of the way that a hooman would. LLMs will only get smarter, and we can only guess what comes after the transformer model anyway. LLMs atm have the very obvious problem of appearing smart but not quite getting all the way there; I think this is pretty much down to it simply trying to predict the next token at its core. Perhaps we'll see something smarter when figure out that an LLM should only be used for output of semantic natural language and not ideation/conceptualisation (which should be handled by a separate model that deals in abstract concepts).
- ric2b 3y agoBut if they first need a browser to chew through the usual tens of megabytes of javascript so that they can finally get the 100KB DOM to parse through, they'll be way less efficient (and often run into the token limits) than if the page has some nice semantic HTML with some optional JS for interactivity (that the AI doesn't need).