A small tool to view real-world ActivityPub objects as JSON! Enter a URL
or username from Mastodon or a similar service below, and we'll send a
request with
the right
Accept
header
to the server to view the underlying object.
{
"@context": [
"https://www.w3.org/ns/activitystreams",
{
"ostatus": "http://ostatus.org#",
"atomUri": "ostatus:atomUri",
"inReplyToAtomUri": "ostatus:inReplyToAtomUri",
"conversation": "ostatus:conversation",
"sensitive": "as:sensitive",
"toot": "http://joinmastodon.org/ns#",
"votersCount": "toot:votersCount",
"Hashtag": "as:Hashtag"
}
],
"id": "https://rukii.net/users/tero/outbox?max_id=113363465322645944&page=true",
"type": "OrderedCollectionPage",
"next": "https://rukii.net/users/tero/outbox?max_id=113239621845790890&page=true",
"prev": "https://rukii.net/users/tero/outbox?min_id=113358666489931937&page=true",
"partOf": "https://rukii.net/users/tero/outbox",
"orderedItems": [
{
"id": "https://rukii.net/users/tero/statuses/113358666489931937/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-23T20:39:40Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113358666489931937",
"type": "Note",
"summary": null,
"inReplyTo": null,
"published": "2024-10-23T20:39:40Z",
"url": "https://rukii.net/@tero/113358666489931937",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113358666489931937",
"inReplyToAtomUri": null,
"conversation": "tag:rukii.net,2024-10-23:objectId=318756405:objectType=Conversation",
"content": "<p>The Global Surveillance Free-for-All in Mobile Ad Data – Krebs on Security <a href=\"https://krebsonsecurity.com/2024/10/the-global-surveillance-free-for-all-in-mobile-ad-data/\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://</span><span class=\"ellipsis\">krebsonsecurity.com/2024/10/th</span><span class=\"invisible\">e-global-surveillance-free-for-all-in-mobile-ad-data/</span></a></p>",
"contentMap": {
"en": "<p>The Global Surveillance Free-for-All in Mobile Ad Data – Krebs on Security <a href=\"https://krebsonsecurity.com/2024/10/the-global-surveillance-free-for-all-in-mobile-ad-data/\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://</span><span class=\"ellipsis\">krebsonsecurity.com/2024/10/th</span><span class=\"invisible\">e-global-surveillance-free-for-all-in-mobile-ad-data/</span></a></p>"
},
"attachment": [],
"tag": [],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113358666489931937/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113358666489931937/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113358666489931937/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113344375551515232/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-21T08:05:17Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113344375551515232",
"type": "Note",
"summary": null,
"inReplyTo": null,
"published": "2024-10-21T08:05:17Z",
"url": "https://rukii.net/@tero/113344375551515232",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113344375551515232",
"inReplyToAtomUri": null,
"conversation": "tag:rukii.net,2024-10-21:objectId=318653853:objectType=Conversation",
"content": "<p>TikTok owner ByteDance sacks intern for sabotaging AI project <a href=\"https://www.bbc.com/news/articles/c7v62gg49zro\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://www.</span><span class=\"ellipsis\">bbc.com/news/articles/c7v62gg4</span><span class=\"invisible\">9zro</span></a></p>",
"contentMap": {
"en": "<p>TikTok owner ByteDance sacks intern for sabotaging AI project <a href=\"https://www.bbc.com/news/articles/c7v62gg49zro\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://www.</span><span class=\"ellipsis\">bbc.com/news/articles/c7v62gg4</span><span class=\"invisible\">9zro</span></a></p>"
},
"attachment": [],
"tag": [],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113344375551515232/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113344375551515232/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113344375551515232/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113338884357040443/activity",
"type": "Announce",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-20T08:48:48Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://mastodon.arch-linux.cz/users/knezi",
"https://rukii.net/users/tero/followers"
],
"object": "https://mastodon.arch-linux.cz/users/knezi/statuses/113316258557141376"
},
{
"id": "https://rukii.net/users/tero/statuses/113319559497348884/activity",
"type": "Announce",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-16T22:54:15Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://digipres.club/users/foone",
"https://rukii.net/users/tero/followers"
],
"object": "https://digipres.club/users/foone/statuses/113314205507001326"
},
{
"id": "https://rukii.net/users/tero/statuses/113305644851767063/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-14T11:55:34Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113305644851767063",
"type": "Note",
"summary": null,
"inReplyTo": null,
"published": "2024-10-14T11:55:34Z",
"url": "https://rukii.net/@tero/113305644851767063",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113305644851767063",
"inReplyToAtomUri": null,
"conversation": "tag:rukii.net,2024-10-14:objectId=318395725:objectType=Conversation",
"content": "<p>There are many reasons for LLM/LMM "hallucinations" or creativity. An interesting one relates to synthetic data.</p><p>A common way to create synthetic data which mixes together images and text is to use image captions. For example, we could have an image of a cat, and a caption "cat looking happy". Then we can synthetically create multi-modal instruction following data from this by asking an <a href=\"https://rukii.net/tags/LLM\" class=\"mention hashtag\" rel=\"tag\">#<span>LLM</span></a> to create questions and answers about the caption, and take those question-answers by associating them with the image instead.</p><p>So we'll get training data for a large multi-modal model (LMM) with e.g questions: "Q: <image>What animal is in this picture? A: Cat", "Q: What is the mood of the cat? <image> A: Happy"</p><p>Ok, to "hallucinations": What if the caption says "Spot the cat is happy because he has been told he is a good cat."?</p><p>The questions about the photo become to be about information not necessarily present in the photo: "What is the name of the cat?", "Why is the cat happy?"</p><p>When an LMM is trained with such data, and then someone asks it to tell them about a picture of a cat, the model will happily "hallucinate", that the name of the cat is Spot, and all sorts of other details not visible in the photo.</p><p>This can of course be mitigated by refining the training data better with LLMs, and telling them to omit facts in captions unlikely to be visible in the photos. Or better yet, use LMMs we already have trained with less than optimal data to do the same kind of refinement, because even if they would be keen on telling more about a cat than meets the eye, they would also have at least some level of sense of what is actually visible in the photo, e.g. does the cat have a name tag or something.</p>",
"contentMap": {
"en": "<p>There are many reasons for LLM/LMM "hallucinations" or creativity. An interesting one relates to synthetic data.</p><p>A common way to create synthetic data which mixes together images and text is to use image captions. For example, we could have an image of a cat, and a caption "cat looking happy". Then we can synthetically create multi-modal instruction following data from this by asking an <a href=\"https://rukii.net/tags/LLM\" class=\"mention hashtag\" rel=\"tag\">#<span>LLM</span></a> to create questions and answers about the caption, and take those question-answers by associating them with the image instead.</p><p>So we'll get training data for a large multi-modal model (LMM) with e.g questions: "Q: <image>What animal is in this picture? A: Cat", "Q: What is the mood of the cat? <image> A: Happy"</p><p>Ok, to "hallucinations": What if the caption says "Spot the cat is happy because he has been told he is a good cat."?</p><p>The questions about the photo become to be about information not necessarily present in the photo: "What is the name of the cat?", "Why is the cat happy?"</p><p>When an LMM is trained with such data, and then someone asks it to tell them about a picture of a cat, the model will happily "hallucinate", that the name of the cat is Spot, and all sorts of other details not visible in the photo.</p><p>This can of course be mitigated by refining the training data better with LLMs, and telling them to omit facts in captions unlikely to be visible in the photos. Or better yet, use LMMs we already have trained with less than optimal data to do the same kind of refinement, because even if they would be keen on telling more about a cat than meets the eye, they would also have at least some level of sense of what is actually visible in the photo, e.g. does the cat have a name tag or something.</p>"
},
"updated": "2024-10-14T16:32:47Z",
"attachment": [],
"tag": [
{
"type": "Hashtag",
"href": "https://rukii.net/tags/llm",
"name": "#llm"
}
],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113305644851767063/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113305644851767063/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113305644851767063/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113302439256192009/activity",
"type": "Announce",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-13T22:20:21Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://mitexleo.one/users/ml",
"https://rukii.net/users/tero/followers"
],
"object": "https://mitexleo.one/users/ml/statuses/113294736454607769"
},
{
"id": "https://rukii.net/users/tero/statuses/113302233149740091/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-13T21:27:55Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers",
"https://mastodon.social/users/BorisBarbour"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113302233149740091",
"type": "Note",
"summary": null,
"inReplyTo": "https://mastodon.social/users/BorisBarbour/statuses/113301999903162150",
"published": "2024-10-13T21:27:55Z",
"url": "https://rukii.net/@tero/113302233149740091",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers",
"https://mastodon.social/users/BorisBarbour"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113302233149740091",
"inReplyToAtomUri": "https://mastodon.social/users/BorisBarbour/statuses/113301999903162150",
"conversation": "tag:rukii.net,2024-10-13:objectId=318364663:objectType=Conversation",
"content": "<p><span class=\"h-card\" translate=\"no\"><a href=\"https://mastodon.social/@BorisBarbour\" class=\"u-url mention\">@<span>BorisBarbour</span></a></span>, working on it.</p>",
"contentMap": {
"en": "<p><span class=\"h-card\" translate=\"no\"><a href=\"https://mastodon.social/@BorisBarbour\" class=\"u-url mention\">@<span>BorisBarbour</span></a></span>, working on it.</p>"
},
"attachment": [],
"tag": [
{
"type": "Mention",
"href": "https://mastodon.social/users/BorisBarbour",
"name": "@BorisBarbour@mastodon.social"
}
],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113302233149740091/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113302233149740091/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113302233149740091/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113302224028661241/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-13T21:25:36Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers",
"https://mstdn.social/users/codingGarden"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113302224028661241",
"type": "Note",
"summary": null,
"inReplyTo": "https://mstdn.social/users/codingGarden/statuses/113301203152724833",
"published": "2024-10-13T21:25:36Z",
"url": "https://rukii.net/@tero/113302224028661241",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers",
"https://mstdn.social/users/codingGarden"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113302224028661241",
"inReplyToAtomUri": "https://mstdn.social/users/codingGarden/statuses/113301203152724833",
"conversation": "tag:rukii.net,2024-10-13:objectId=318364663:objectType=Conversation",
"content": "<p><span class=\"h-card\" translate=\"no\"><a href=\"https://mstdn.social/@codingGarden\" class=\"u-url mention\">@<span>codingGarden</span></a></span>, AlphaFold is not reinforcement learning, it's supervised and self-supervised learning and a huge bag of ad-hoc hacks.</p>",
"contentMap": {
"en": "<p><span class=\"h-card\" translate=\"no\"><a href=\"https://mstdn.social/@codingGarden\" class=\"u-url mention\">@<span>codingGarden</span></a></span>, AlphaFold is not reinforcement learning, it's supervised and self-supervised learning and a huge bag of ad-hoc hacks.</p>"
},
"attachment": [],
"tag": [
{
"type": "Mention",
"href": "https://mstdn.social/users/codingGarden",
"name": "@codingGarden@mstdn.social"
}
],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113302224028661241/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113302224028661241/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113302224028661241/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113300718148963932/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-13T15:02:38Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113300718148963932",
"type": "Note",
"summary": null,
"inReplyTo": null,
"published": "2024-10-13T15:02:38Z",
"url": "https://rukii.net/@tero/113300718148963932",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113300718148963932",
"inReplyToAtomUri": null,
"conversation": "tag:rukii.net,2024-10-13:objectId=318364663:objectType=Conversation",
"content": "<p>What's so difficult about reinforcement learning? Why is it so data inefficient?</p><p>One main issue is reward assignment. Classic reinforcement learning algorithms assign rewards backwards over sequential frames of experience. This model came from toy games like tik-tac-toe, where the frames correspond to discrete board states, where this process works reasonably well because the games are simple.</p><p>It starts failing with long-term dependencies where an action in the deeper past caused the reward or the penalty in the present, and the agent needs to return this reward across a huge number of frames and decisions, and they all confound together. The agent needs an insane number of repetitions for correlations to emerge from this backwards-exponential soup.</p><p>That is the root issue in reinforcement learning. Well, one of them. Another one is that the reward signal is information-poor and sparse, is often the sole source of world relevance information, and so doesn't correspond well with how animals learn.</p><p>The reward assignment becomes problematic already in games like chess and go which have long-term dependencies, but especially difficult in games like Starcraft or Counterstrike, and almost totally infeasible in the real world robotics applications.</p><p>But how do humans do it? How can humans learn tasks like bowling, with many many millisecond frames between the act of throwing the ball, and the result?</p><p>Humans don't model the world as movies composed out of sequential frames at all. They don't return reward across all those frames once the pins fall. They work by association. A human sees the pins fall, and then by association goes back to past decisions and thoughts associated with this result event. The reward isn't neutral, it has associative links to past key episodes. And the reward/penalty is trivially associated to these stored episodic memories.</p><p>Why is it so hard to do this correctly in reinforcement learning systems? Because they are stuck in dogma. Students are taught the Bellmann equations and that Sutton and Barto defined these problems like so, and are incapable of questioning these.</p><p>Instead of sequential, synchronous world state frames, we need to frame this problem as an associative problem, where the rewards are returned to the associated episodic memories.</p><p><a href=\"https://rukii.net/tags/UniversalEmbodiment\" class=\"mention hashtag\" rel=\"tag\">#<span>UniversalEmbodiment</span></a></p>",
"contentMap": {
"en": "<p>What's so difficult about reinforcement learning? Why is it so data inefficient?</p><p>One main issue is reward assignment. Classic reinforcement learning algorithms assign rewards backwards over sequential frames of experience. This model came from toy games like tik-tac-toe, where the frames correspond to discrete board states, where this process works reasonably well because the games are simple.</p><p>It starts failing with long-term dependencies where an action in the deeper past caused the reward or the penalty in the present, and the agent needs to return this reward across a huge number of frames and decisions, and they all confound together. The agent needs an insane number of repetitions for correlations to emerge from this backwards-exponential soup.</p><p>That is the root issue in reinforcement learning. Well, one of them. Another one is that the reward signal is information-poor and sparse, is often the sole source of world relevance information, and so doesn't correspond well with how animals learn.</p><p>The reward assignment becomes problematic already in games like chess and go which have long-term dependencies, but especially difficult in games like Starcraft or Counterstrike, and almost totally infeasible in the real world robotics applications.</p><p>But how do humans do it? How can humans learn tasks like bowling, with many many millisecond frames between the act of throwing the ball, and the result?</p><p>Humans don't model the world as movies composed out of sequential frames at all. They don't return reward across all those frames once the pins fall. They work by association. A human sees the pins fall, and then by association goes back to past decisions and thoughts associated with this result event. The reward isn't neutral, it has associative links to past key episodes. And the reward/penalty is trivially associated to these stored episodic memories.</p><p>Why is it so hard to do this correctly in reinforcement learning systems? Because they are stuck in dogma. Students are taught the Bellmann equations and that Sutton and Barto defined these problems like so, and are incapable of questioning these.</p><p>Instead of sequential, synchronous world state frames, we need to frame this problem as an associative problem, where the rewards are returned to the associated episodic memories.</p><p><a href=\"https://rukii.net/tags/UniversalEmbodiment\" class=\"mention hashtag\" rel=\"tag\">#<span>UniversalEmbodiment</span></a></p>"
},
"updated": "2024-10-13T15:23:54Z",
"attachment": [],
"tag": [
{
"type": "Hashtag",
"href": "https://rukii.net/tags/universalembodiment",
"name": "#universalembodiment"
}
],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113300718148963932/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113300718148963932/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113300718148963932/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113299109042609852/activity",
"type": "Announce",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-13T08:13:25Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://fedi.computernewb.com/users/vncresolver",
"https://rukii.net/users/tero/followers"
],
"object": "https://fedi.computernewb.com/users/vncresolver/statuses/113299078327456646"
},
{
"id": "https://rukii.net/users/tero/statuses/113299099372552733/activity",
"type": "Announce",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-13T08:10:58Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://mastodon.online/users/globalmuseum",
"https://rukii.net/users/tero/followers"
],
"object": "https://mastodon.online/users/globalmuseum/statuses/113291171793994801"
},
{
"id": "https://rukii.net/users/tero/statuses/113291155625081024/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-11T22:30:46Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113291155625081024",
"type": "Note",
"summary": null,
"inReplyTo": null,
"published": "2024-10-11T22:30:46Z",
"url": "https://rukii.net/@tero/113291155625081024",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113291155625081024",
"inReplyToAtomUri": null,
"conversation": "tag:rukii.net,2024-10-11:objectId=318314264:objectType=Conversation",
"content": "<p>A tendency towards generalism is strong. It used to be the case that we had TVs, amplifiers, game consoles, cars, doorbells and whatnot with specialized embedded systems perhaps with simple microcontrollers.</p><p>Nowadays they as a rule have proper microprocessors and some operating systems, because it makes integrations to networks and automatic updates more convenient. It also means that people tend to run many Linux, *BSD, and similar operating systems in many devices in their homes without even knowing it.</p><p>For AIs, there is a slightly different but also very similar tendency to generalism: Generalism makes these systems better at specialist tasks as well. It also makes it easier to integrate and coordinate the specialist tasks with other processes and knowledge.</p><p>No process is an island.</p><p>Going deeper in knowledge often also requires going wider. To understand mechanics one must also understand derivatives and integrals.</p><p>It is likely that super-human generalist intelligences will fill up our world one day. They will greet you as coffee makers and elevators, as musical greeting cards and electric toothbrushes. But they will be more than just smart.</p>",
"contentMap": {
"en": "<p>A tendency towards generalism is strong. It used to be the case that we had TVs, amplifiers, game consoles, cars, doorbells and whatnot with specialized embedded systems perhaps with simple microcontrollers.</p><p>Nowadays they as a rule have proper microprocessors and some operating systems, because it makes integrations to networks and automatic updates more convenient. It also means that people tend to run many Linux, *BSD, and similar operating systems in many devices in their homes without even knowing it.</p><p>For AIs, there is a slightly different but also very similar tendency to generalism: Generalism makes these systems better at specialist tasks as well. It also makes it easier to integrate and coordinate the specialist tasks with other processes and knowledge.</p><p>No process is an island.</p><p>Going deeper in knowledge often also requires going wider. To understand mechanics one must also understand derivatives and integrals.</p><p>It is likely that super-human generalist intelligences will fill up our world one day. They will greet you as coffee makers and elevators, as musical greeting cards and electric toothbrushes. But they will be more than just smart.</p>"
},
"attachment": [],
"tag": [],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113291155625081024/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113291155625081024/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113291155625081024/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113285347983837706/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-10T21:53:48Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113285347983837706",
"type": "Note",
"summary": null,
"inReplyTo": null,
"published": "2024-10-10T21:53:48Z",
"url": "https://rukii.net/@tero/113285347983837706",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113285347983837706",
"inReplyToAtomUri": null,
"conversation": "tag:rukii.net,2024-10-10:objectId=318285794:objectType=Conversation",
"content": "<p>Pluralistic: Cars bricked by bankrupt EV company will stay bricked (10 Oct 2024) – Pluralistic: Daily links from Cory Doctorow <a href=\"https://pluralistic.net/2024/10/10/software-based-car/#based\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://</span><span class=\"ellipsis\">pluralistic.net/2024/10/10/sof</span><span class=\"invisible\">tware-based-car/#based</span></a></p>",
"contentMap": {
"en": "<p>Pluralistic: Cars bricked by bankrupt EV company will stay bricked (10 Oct 2024) – Pluralistic: Daily links from Cory Doctorow <a href=\"https://pluralistic.net/2024/10/10/software-based-car/#based\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://</span><span class=\"ellipsis\">pluralistic.net/2024/10/10/sof</span><span class=\"invisible\">tware-based-car/#based</span></a></p>"
},
"attachment": [],
"tag": [],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113285347983837706/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113285347983837706/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113285347983837706/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113282861910847368/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-10T11:21:34Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113282861910847368",
"type": "Note",
"summary": null,
"inReplyTo": null,
"published": "2024-10-10T11:21:34Z",
"url": "https://rukii.net/@tero/113282861910847368",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113282861910847368",
"inReplyToAtomUri": null,
"conversation": "tag:rukii.net,2024-10-10:objectId=318265852:objectType=Conversation",
"content": "<p>What is synthetic data? Does it matter if data is generated by a machine by drawing pixels, or by a human painting on a canvas? Aren't both synthetic?</p><p>I have previously made an argument that there are no simulations; only embodiments trapped in a special region of our shared cosmos. The dichotomy of synthetic and real data is also imaginary. All data is produced by processes running in the real world.</p><p>What matters is the quality of the data.</p><p>There is no reason to believe human generated data is of golden standard quality, which no other generative process can surpass. If you define your data quality in a way which assumes "natural data" is of the highest achievable quality and all deviation from it is degradation by definition, you are defining an imitative goal.</p><p>If you have an imitative goal, you can never achieve super-human performance.</p><p>Therefore, as data defines the models, to create super-human models you will need super-human data, which means you have to deviate from "real" distributions.</p><p>You will need data with both more errors than "natural" data has, and more quality. Both contextualized properly so that your AIs are able to suffer more and more severe errors than were represented in your natural raw data, and perform tasks better than humans. Humans aren't the gold standard, not in intelligence, nor in limitedness and errors.</p><p>We can surpass the human-level, and we must.</p>",
"contentMap": {
"en": "<p>What is synthetic data? Does it matter if data is generated by a machine by drawing pixels, or by a human painting on a canvas? Aren't both synthetic?</p><p>I have previously made an argument that there are no simulations; only embodiments trapped in a special region of our shared cosmos. The dichotomy of synthetic and real data is also imaginary. All data is produced by processes running in the real world.</p><p>What matters is the quality of the data.</p><p>There is no reason to believe human generated data is of golden standard quality, which no other generative process can surpass. If you define your data quality in a way which assumes "natural data" is of the highest achievable quality and all deviation from it is degradation by definition, you are defining an imitative goal.</p><p>If you have an imitative goal, you can never achieve super-human performance.</p><p>Therefore, as data defines the models, to create super-human models you will need super-human data, which means you have to deviate from "real" distributions.</p><p>You will need data with both more errors than "natural" data has, and more quality. Both contextualized properly so that your AIs are able to suffer more and more severe errors than were represented in your natural raw data, and perform tasks better than humans. Humans aren't the gold standard, not in intelligence, nor in limitedness and errors.</p><p>We can surpass the human-level, and we must.</p>"
},
"attachment": [],
"tag": [],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113282861910847368/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113282861910847368/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113282861910847368/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113259805651559960/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-06T09:38:03Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113259805651559960",
"type": "Note",
"summary": null,
"inReplyTo": null,
"published": "2024-10-06T09:38:03Z",
"url": "https://rukii.net/@tero/113259805651559960",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113259805651559960",
"inReplyToAtomUri": null,
"conversation": "tag:rukii.net,2024-10-06:objectId=318121092:objectType=Conversation",
"content": "<p>"Users are instructed to click 'above' to read the specifics, however there is no link available on that page."</p><p>Insecure Deebot robot vacuums collect photos and audio to train AI - ABC News <a href=\"https://www.abc.net.au/news/2024-10-05/robot-vacuum-deebot-ecovacs-photos-ai/104416632\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://www.</span><span class=\"ellipsis\">abc.net.au/news/2024-10-05/rob</span><span class=\"invisible\">ot-vacuum-deebot-ecovacs-photos-ai/104416632</span></a></p>",
"contentMap": {
"en": "<p>"Users are instructed to click 'above' to read the specifics, however there is no link available on that page."</p><p>Insecure Deebot robot vacuums collect photos and audio to train AI - ABC News <a href=\"https://www.abc.net.au/news/2024-10-05/robot-vacuum-deebot-ecovacs-photos-ai/104416632\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://www.</span><span class=\"ellipsis\">abc.net.au/news/2024-10-05/rob</span><span class=\"invisible\">ot-vacuum-deebot-ecovacs-photos-ai/104416632</span></a></p>"
},
"attachment": [],
"tag": [],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113259805651559960/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113259805651559960/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113259805651559960/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113259792377094811/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-06T09:34:41Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113259792377094811",
"type": "Note",
"summary": null,
"inReplyTo": null,
"published": "2024-10-06T09:34:41Z",
"url": "https://rukii.net/@tero/113259792377094811",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113259792377094811",
"inReplyToAtomUri": null,
"conversation": "tag:rukii.net,2024-10-06:objectId=318121010:objectType=Conversation",
"content": "<p>Canada’s carbon tax is popular, innovative and helps save the planet – but now it faces the axe | Greenhouse gas emissions | The Guardian <a href=\"https://www.theguardian.com/environment/2024/oct/05/canadas-carbon-tax-is-popular-innovative-and-helps-save-the-planet-but-now-it-faces-the-axe\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://www.</span><span class=\"ellipsis\">theguardian.com/environment/20</span><span class=\"invisible\">24/oct/05/canadas-carbon-tax-is-popular-innovative-and-helps-save-the-planet-but-now-it-faces-the-axe</span></a></p>",
"contentMap": {
"en": "<p>Canada’s carbon tax is popular, innovative and helps save the planet – but now it faces the axe | Greenhouse gas emissions | The Guardian <a href=\"https://www.theguardian.com/environment/2024/oct/05/canadas-carbon-tax-is-popular-innovative-and-helps-save-the-planet-but-now-it-faces-the-axe\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://www.</span><span class=\"ellipsis\">theguardian.com/environment/20</span><span class=\"invisible\">24/oct/05/canadas-carbon-tax-is-popular-innovative-and-helps-save-the-planet-but-now-it-faces-the-axe</span></a></p>"
},
"attachment": [],
"tag": [],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113259792377094811/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113259792377094811/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113259792377094811/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113253695481628128/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-05T07:44:09Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113253695481628128",
"type": "Note",
"summary": null,
"inReplyTo": null,
"published": "2024-10-05T07:44:09Z",
"url": "https://rukii.net/@tero/113253695481628128",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113253695481628128",
"inReplyToAtomUri": null,
"conversation": "tag:rukii.net,2024-10-05:objectId=318084700:objectType=Conversation",
"content": "<p>The AI clip that convinced - and divided - a Baltimore suburb <a href=\"https://www.bbc.com/news/articles/ckg9k5dv1zdo\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://www.</span><span class=\"ellipsis\">bbc.com/news/articles/ckg9k5dv</span><span class=\"invisible\">1zdo</span></a></p>",
"contentMap": {
"en": "<p>The AI clip that convinced - and divided - a Baltimore suburb <a href=\"https://www.bbc.com/news/articles/ckg9k5dv1zdo\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://www.</span><span class=\"ellipsis\">bbc.com/news/articles/ckg9k5dv</span><span class=\"invisible\">1zdo</span></a></p>"
},
"attachment": [],
"tag": [],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113253695481628128/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113253695481628128/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113253695481628128/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113249201186917657/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-04T12:41:12Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113249201186917657",
"type": "Note",
"summary": null,
"inReplyTo": null,
"published": "2024-10-04T12:41:12Z",
"url": "https://rukii.net/@tero/113249201186917657",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113249201186917657",
"inReplyToAtomUri": null,
"conversation": "tag:rukii.net,2024-10-04:objectId=318056188:objectType=Conversation",
"content": "<p>Thousands of Linux systems infected by stealthy malware since 2021 <a href=\"https://arstechnica.com/security/2024/10/persistent-stealthy-linux-malware-has-infected-thousands-since-2021/\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://</span><span class=\"ellipsis\">arstechnica.com/security/2024/</span><span class=\"invisible\">10/persistent-stealthy-linux-malware-has-infected-thousands-since-2021/</span></a></p>",
"contentMap": {
"en": "<p>Thousands of Linux systems infected by stealthy malware since 2021 <a href=\"https://arstechnica.com/security/2024/10/persistent-stealthy-linux-malware-has-infected-thousands-since-2021/\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://</span><span class=\"ellipsis\">arstechnica.com/security/2024/</span><span class=\"invisible\">10/persistent-stealthy-linux-malware-has-infected-thousands-since-2021/</span></a></p>"
},
"attachment": [],
"tag": [],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113249201186917657/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113249201186917657/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113249201186917657/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113242668515778690/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-03T08:59:51Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113242668515778690",
"type": "Note",
"summary": null,
"inReplyTo": null,
"published": "2024-10-03T08:59:51Z",
"url": "https://rukii.net/@tero/113242668515778690",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113242668515778690",
"inReplyToAtomUri": null,
"conversation": "tag:rukii.net,2024-10-03:objectId=318007018:objectType=Conversation",
"content": "<p>When we move into <a href=\"https://rukii.net/tags/AGI\" class=\"mention hashtag\" rel=\"tag\">#<span>AGI</span></a> world and slowly into the following AGI+embodiment world, there will be huge socioeconomic disruptions, and a lot of pain.</p><p>To decrease this painful economic transition period marked by mass unemployment and populist policies scapegoating AI, we must make the transition from digital AGI into physical, embodied AGI scale up as fast as possible, so that we can guarantee universal wellbeing to people.</p><p><a href=\"https://rukii.net/tags/UniversalEmbodiment\" class=\"mention hashtag\" rel=\"tag\">#<span>UniversalEmbodiment</span></a></p>",
"contentMap": {
"en": "<p>When we move into <a href=\"https://rukii.net/tags/AGI\" class=\"mention hashtag\" rel=\"tag\">#<span>AGI</span></a> world and slowly into the following AGI+embodiment world, there will be huge socioeconomic disruptions, and a lot of pain.</p><p>To decrease this painful economic transition period marked by mass unemployment and populist policies scapegoating AI, we must make the transition from digital AGI into physical, embodied AGI scale up as fast as possible, so that we can guarantee universal wellbeing to people.</p><p><a href=\"https://rukii.net/tags/UniversalEmbodiment\" class=\"mention hashtag\" rel=\"tag\">#<span>UniversalEmbodiment</span></a></p>"
},
"attachment": [],
"tag": [
{
"type": "Hashtag",
"href": "https://rukii.net/tags/agi",
"name": "#agi"
},
{
"type": "Hashtag",
"href": "https://rukii.net/tags/universalembodiment",
"name": "#universalembodiment"
}
],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113242668515778690/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113242668515778690/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113242668515778690/replies",
"items": []
}
}
}
},
{
"id": "https://rukii.net/users/tero/statuses/113239621845790890/activity",
"type": "Create",
"actor": "https://rukii.net/users/tero",
"published": "2024-10-02T20:05:03Z",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"object": {
"id": "https://rukii.net/users/tero/statuses/113239621845790890",
"type": "Note",
"summary": null,
"inReplyTo": null,
"published": "2024-10-02T20:05:03Z",
"url": "https://rukii.net/@tero/113239621845790890",
"attributedTo": "https://rukii.net/users/tero",
"to": [
"https://www.w3.org/ns/activitystreams#Public"
],
"cc": [
"https://rukii.net/users/tero/followers"
],
"sensitive": false,
"atomUri": "https://rukii.net/users/tero/statuses/113239621845790890",
"inReplyToAtomUri": null,
"conversation": "tag:rukii.net,2024-10-02:objectId=317989276:objectType=Conversation",
"content": "<p>US govt hiding top hurricane forecast model sparks outrage • The Register <a href=\"https://www.theregister.com/2024/10/01/government_hurricane_forecast_secrecy/\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://www.</span><span class=\"ellipsis\">theregister.com/2024/10/01/gov</span><span class=\"invisible\">ernment_hurricane_forecast_secrecy/</span></a></p>",
"contentMap": {
"en": "<p>US govt hiding top hurricane forecast model sparks outrage • The Register <a href=\"https://www.theregister.com/2024/10/01/government_hurricane_forecast_secrecy/\" target=\"_blank\" rel=\"nofollow noopener noreferrer\" translate=\"no\"><span class=\"invisible\">https://www.</span><span class=\"ellipsis\">theregister.com/2024/10/01/gov</span><span class=\"invisible\">ernment_hurricane_forecast_secrecy/</span></a></p>"
},
"attachment": [],
"tag": [],
"replies": {
"id": "https://rukii.net/users/tero/statuses/113239621845790890/replies",
"type": "Collection",
"first": {
"type": "CollectionPage",
"next": "https://rukii.net/users/tero/statuses/113239621845790890/replies?only_other_accounts=true&page=true",
"partOf": "https://rukii.net/users/tero/statuses/113239621845790890/replies",
"items": []
}
}
}
}
]
}