would it be possible if the TASer wants to for the movie publish main encode being the one in his own channel? this might give more appeal for TASers who didn't submit for lack of redirection for their own youtube channel and could remove the need for site encodes
example: current darkman submission for killer instinct gameboy has a great encode and he could theoretically request his encode to be the main one the site directs to, the tas encode still can exist normally but could even speed up publish process if not needed a site encode
it's only an example, i don't think darkman wants that specifically
Site archive encodes are not going away. However, the option for submitters to opt for the publication encode to be their own is planned.
That said, submitters are already allowed to request that their encode be the one used on the publication. There just isn't anything explicitly in the submission form to indicate that yet.
I tried to estimate the cost of AI TAS slop, and I keep getting less than $1000 per 3 minutes of manual tasing of a simple game at my optimization level, assuming the AI vibe taser is not smart or experienced. A smarter and more experienced AI vibe taser can maybe come close to a cost of regular botting with a computer + 1% of the cost of running some AI model. The AI model does not need to run continuously, so it can idle and wait for a regular bot to finish a task. The AI model can run for a minute using the data center, then the regular bot works for 99 minutes using a laptop. There is no reliable estimate of the cost because nobody has optimized the process.
Wait what? Nothing here makes any sense. Manual TASing costs practically nothing. It's hard to measure precisely how much computer wattage it takes to run an emulator on a typical computer, but even if we take a rough overestimate of 500W, with a rate of $0.15/kWh, it'd cost about $0.075 per hour. Even if there were 1000 people TASing continuously, the cost would pale in comparison to that of AI.
The cost of using AI to TAS would vary wildly depending on how AI is being used. There's a ton of different methodologies that might be employed, and all of them may be run locally or via an AI service. If done via a service, the only measurable cost is token usage charged by the service, but it's likely they are undercharging right now to get people hooked on their services. Plus none of that factors in the cost to train those AI models. That cost cannot be simply calculated by any of us (and frankly I doubt the AI companies even know precisely how much they spend on AI training). Not to mention the secondary costs like the massive price increases for RAM and flash memory.
alexheights1 wrote:
I would lower the max submission count per week to 1 or even less than that. The submission limit has almost no effect on anyone but can reduce the number of AI TAS slop submissions. If a trusted user can give a reasonable argument for why they need to be able to post 3 submissions per week, then give them special permission.
Why are you talking about lowering the submission limit? What AI TAS slop submissions are we trying to reduce the rate of?? If we had a problem with submitting AI slop, we would take more direct actions to prevent it, not just slow down the rate that they'd submit it.
We're not adding Chimera. Not because it may have more bugs than bizhawk. Not because it's more or less stable than bizhawk. Not because it might have bad accuracy. Not because it's new, regardless if it may get better over time. Those points are irrelevant here, and in the case of new cores, wouldn't necessarily prevent an emulator from being accepted anyways.
As Samsara has both mentioned in general and alluded to, and as I argued yesterday, Chimera's development heavily uses generative AI which directly goes against our stance as outlined in the site's AI Policy.
(Before someone rule-lawyers with me that the policy doesn't specifically outlaw AI-developed emulators: Yes, we know it doesn't. But Chimera's development so overwhelmingly uses AI, accepting it would fly right in the face of the policy's spirit and intent.)
Can TASVideos support Chimera movie and tasstudio projects? I can't upload the userfile here and can't submit.
KusogeMan wrote:
I'll just post Chimera movies elsewhere if that specific emulator isn't supported or videos only for the forum.
However the emulator is progressing well and it feels more like an opportunity to have it, i enjoyed a little bit TASing with it and this a thread about TASing options.
Considering we just recently enacted an AI Policy in response to our commitment to reject GenAI where reasonable possible, I find it extremely unlikely we'd ever support an emulator that is so heavily developed using AI as Chimera is. I'd personally be against accepting it for sure. Not to mention that its primary (human) developer was banned for bigotry and utter disregard for human life.
You are of course free to post your Chimera movies elsewhere if you so choose.
Do you have any kind of evidence to support what you're suggesting?
This is a great question. Evidence to verify/refute my Herculean deduction exists, but we have no access to it:
The swarm produced ~4.9 million messages and ~300 billion output tokens, with ~2.7 million messages and ~130 billion output tokens on Navier–Stokes specifically.
We do not have access to this data, but my guess is that TASVideos is mentioned somewhere in the millions of messages.
So you don't actually know any of this for a fact. It's reasonable to assume that various AI crawlers have scanned TASVideos, however, it's not reasonable to assume or conclude that AI must now be capable of producing fully working (let alone optimized) TASes. Doesn't matter how many AI agents or tokens have been produced, that doesn't inherently mean anything in regards to TAS production.
alexheights1 wrote:
If there is some confusion in my statements it is due to the complexity of the topic. There are new kinds of ways to do things and it is really difficult to understand what is what. For example, the AI policy itself is not completely clear about the difference between neural network and generative AI. It is really difficult to write about the topic perfectly.
The topic is indeed complex, but it's very unclear what your intent is with your recent posts. What arguments are you trying to convey? What purpose do your posts have? How do they contribute to the topic?
Here's a summary of the basic concepts at play here:
A neural network is fundamentally a weighted decision graph inspired by how neurons work in a living brain. This concept in machine learning has existed far before generative AI or LLMs have existed and by themselves are not a inherently an issue.
Large Language Models (LLMs) are massive computer models (typically in the form of a neural network) trained on massive quantities of human language in order to predict the next word/token in a string of text. What makes LLMs concerning is their size and the computational cost of both training and using them, and all subsequent costs. LLMs are a form of generative AI which produces text given some input.
Generative AI is an umbrella term for artificial intelligence that takes some input and produces new data as a result. This is commonly either text, images/video, or audio but isn't limited to just that. Before GenAI exploded, most neural network AIs were used to classify data or make simple decisions like identifying if an image contains a particular kind of object. They didn't generate entirely new images.
alexheights1 wrote:
I support the AI policy, but I have not seen any real data where the environmental effects of a manual TAS are compared to a similar AI-generated TAS, so the idea that AI-generated input is environmentally detrimental is based on my uneducated guess. Maybe my manual tases are worse for the environment compared to generating the input with AI? I don't know.
There's no conceivable way your manual TASes are anywhere near the costs of training or using generative AI. I'm sure you can find some data if you try researching the topic. Try running a generative model on your local computer and monitor the time and resources it takes (that's what I have done). It's noticeably larger than running a TASing emulator, even when utilizing scripts. Then consider that the training of these models takes drastically more power than simply using it to generate output.
Not to mention that there are more costs than just computation of the model itself. Scraping the web for training data inherently adds cost at every stage of the communication between the crawler and the web server. Then there's also the ethical problems of deriving new content using the works of others, all without attribution or observing licensing.
alexheights1 wrote:
but what can I do about it?
I don't know, you're the one suddenly talking about it despite the AI Policy having been in place for over a month now. It's very confusing what you are trying to convey here.
alexheights1 wrote:
There is no method for telling if some input is made with LLM/AI or not. This problem exists in many online art communities, and as far as I know there is no solution for it.
There may be some tells for what input is made using GenAI/LLMs just as there are tells for AI generated text. Hard to know for sure at this time. As for knowing the problem exists, yeah we already know that.
They probably went and told every colleague (the other 9,999 AI-agents) that they found this information on TASVideos forum (in the cached internet version). The AI-agents would then calculate that TASVideos may be very valuable source when trying to solve the Navier-Stokes problem. The AI-agents may have then studied TASVideos carefully, learning the principles of tasing from the wiki. They may have then used some tasing principles to find input that gave a solution that they were looking for. The Navier-Stokes problem is about fluid simulations. The question is essentially: Does there exist some input that gives infinite velocity? The answer turned out to be yes, and the 10,000 AI-agents working for 88 hours found the answer. The whole thing is in many ways similar to TAS.
There sure are a lot of maybe's in these statements. Do you have any kind of evidence to support what you're suggesting?
alexheights1 wrote:
I think this is impossible to verify, so it is kind of an useless statement? If somebody makes TAS input with AI, why would they reveal how they did it? They will just submit a TAS and not reveal the AI-usage. And people will never know.
Are you trying to suggest that we shouldn't have bothered to implement an AI policy? If not, then what point are you trying to make?
alexheights1 wrote:
If GPT-6 Astra codes a bot for tasing from scratch and then the human (or GPT-6 Astra) runs the tasbot.exe program which generates the input, would this be allowed? In this case the input itself is not made with generative AI directly, but the program that prints the input is made with generative AI.
Currently the AI Policy states: "AI-generated scripts and tools may still be used in the process of creating an input file as long as the input itself is free of generative AI." (emphasis added)
If someone used AI to produce a program where the inputs were already decided by the AI, then I'd say that breaks the policy. Additionally, using AI to create a tool that itself uses AI to generate inputs is equally problematic.
The spirit of the policy's wording is to allow the use of AI in the creation of tools that humans can use to aid in the development of a TAS. Some example may include calculators, brute force scripts, automation, memory scanners, and more. But tools that use LLMs to make decisions of what inputs to use would be going too far in my opinion.
TASBot is solely a mascot of the TASBot community. He is not something you can purchase or replicate yourself.
TASBot doesn't do any kind of TAS replay himself, rather, at times one of any number of replay devices is setup on/nearby him which is what actually performs the TAS replays on console.
I understand. This was Duck's first submission though, so I feel like it'd be best to allow them to get better at TASing and submit a run they made.
i fear similarities to KotyaTAS's situation, if that was the case though
The situation with KotyaTAS is entirely different. KotyaTAS is not banned, and their issue was repeatedly submitting blatantly unoptimized work. The work was entirely their own though.
Duck not only submitted someone else's work without permission, they also falsely claimed to have contributed inputs to that work (aka plagiarism). Stealing others' work is absolutely unacceptable!
DJ_Incendration wrote:
I understand. This was Duck's first submission though, so I feel like it'd be best to allow them to get better at TASing and submit a run they made.
This is a warning. Continued bad faith arguing, or any other rule-breaking, will result in a ban.
Well this new TASVideos adminstration really likes to ban people, just for making arguments. I've been banned in 2023 just for defending my work. You guys really show no improvement. This abuse of power makes me sick.
Who are you to call me with "bad faith"? You're a mind reader who can read my faith? I'm here to raise a real issue.
It's really funny to me, that instead of defending the published SMB TAS and clearing up the confusion, you guys seem to be more concerned about donating to a former RTA world record holder of SMB, who makes a lot more money in a single livestream than what I made from TASing in 20 years.
We (the moderation team and senior staff) do not ban people for simply making arguments. You were banned in 2023 for repeatedly attacking a staff member under the claim that she intentionally changed the site rules to spite you and your submission, despite being warned to act civil. You were then warned in late-2024 for attacking a different staff member.
Your actions today have not been mere arguments. Despite the warning you were just given, rather than reflecting on your behavior, you have chosen to escalate with more baseless claims against us; all while continuing to assert false narratives about our decisions and your own past history.
It's clear your behavior has not improved since your previous ban. As such, you are now banned indefinitely.
All your talk about "not drive them away", and I see no guilt in you actually driving talented TASers away.
The only people we intentionally "drive away" are those who are actively harmful to this community or other individuals.
HappyLee wrote:
Bigbass wrote:
Separately, if you have a problem with speech from other communities, you should bring that up with them, not us.
I will. I raised this issue here because:
I noticed that the most prominent item on the TASVideos homepage this week has been “Help LeKukie Fix His Roof.”
LeKukie wasn't even an active TASVideos user. He hasn't spoken on TASVideos for 6 years. He's an active member of the RTA community.
I'm sorry you have a problem with us supporting another human being, who is in fact a TASer, and has undoubtedly contributed to the TASing scene.
It is all the more apparent that this thread has been created and argued by you in bad faith. You're not here to simply clear up some confusion or amicably discuss our support for alternative timing methods. You're here to complain that we are using our power of being a community to help people in need, as well as to create false narratives about our own site changes and policies. All under the guise of "stand[ing] together" in support of TASing.
Stop asserting that we made the alt. timing changes because of the RTA community or that we are adhering to RTA rules. That is outright false. We have traditionally already accepted movies optimized for alternative timing methods; the only change we made recently was to explicitly recognize which method was used.
This is a warning. Continued bad faith arguing, or any other rule-breaking, will result in a ban.
Only using TAS timing will not stop people making TASes using other timing methods or listing them as such on YouTube or other sites - it just means they're less likely to submit their work here, which I see as a loss.
That I don't agree. We have traditions and standards that has been held for more than 20 years, and there's nothing wrong with that.
Some of our past traditions and standards have been outright harmful towards the community and to our own staff members. Some have caused a sense of elitism or exclusion, further driving TASers away. Blindly following tradition for the sake of tradition is not a wise choice.
HappyLee wrote:
You said "TASVideos is not an authority on TASing", but nearly all of the world's best TASers are here, and we should represent the TAS community.
Whether you agree or not, that is objectively false. And, if we are to represent "the TAS community" then we MUST take into account and respect that there are major TASing groups who, for good reasons, do not always time their TASes from power-on.
HappyLee wrote:
Walgrey wrote:
How could that be anything but a loss? We want people to feel free to submit their TASes here, not drive them away.
They are free to submit their TASes here, but we should upheld our timing standards. It's been well-known that TAS timing starts at power-on.
First of all, no it's not well-known, and second, we ARE upholding our timing standards. If anything, we are upholding them even more than we already were by explicitly stating where the stated time comes from in the title, and codifying those different timing methodologies.
We have long already judged, accepted, and obsoleted movies based on alternative timing methods, such as in-game time. The issue is that once published, we would display the traditional TAS time (from power on) which sometimes led to very confusing and misleading situations; such as where a faster movie listed a time that was longer than the movie it obsoleted.
HappyLee wrote:
If anyone doesn't agree that TAS should use TAS timing and not RTA timing, then they shouldn't submit their work here. I don't see a loss in that.
Alright, then I shall use your own logic: If you can't agree with all of us to acknowledge and respect alternative timing methods and the TAS communities that actively use those methods for TASing, then you are free to not submit your work here. I don't see a loss in that.
Separately, if you have a problem with speech from other communities, you should bring that up with them, not us.
Hello! Just following up on my submission. I was wondering if there are any updates regarding the judging. I understand that judging takes time, and I appreciate your work. Please let me know if you need anything from me. Thanks!
All submissions have a 72-hour minimum period of time before judging can take place. As of the time of this post, there are still 31 hours remaining (you can check this on the submission page, just above the status header.)
However, as Spikestuff has already noted earlier, this submission is significantly slower than the currently published TAS in this category. So I suspect that it'll be rejected for insufficient optimization.
Please take the time to research and study the resources available on our wiki and elsewhere, as well as looking through the previously published TASes for this game (or for any other you may be interested in). If you only want feedback on your TASes, you should consider uploading to your UserFiles instead of using the submission process, and share that on our forums or Discord. That said, if you do choose to submit TASes for publication again in the future, please be sure you read and understand our submission instructions first.
Usage of LLM-assisted machine translation tools for site communication and accessibility is allowed as long as the original untranslated text is provided.
I don't suppose linking back to the original source would be an option here, would it?
Linking to the original text, while perhaps convenient, isn't a good option. The linked content might change over time (e.g. a wiki article), or might disappear altogether. Having the original on the site ensures that it won't become lost. Plus, it may help deter those who would abuse translation as an excuse/avenue for using generative AI to create forum posts.
Asumeh wrote:
Also, there exists voice-typing/speech-to-text tools for outputting text with capitalization and punctuation through your voice, such as Handy and OpenWhispr, in case manual typing happens to become a burden. I'd assume usage of those tools must be disclosed or perhaps be restricted to a minimum as well.
Voice recognition has existed for decades, even Windows XP had it built in (it wasn't very good, but that was 25 years ago.) Most tools like this will have used some form of neural network. The difference nowadays is that it's not just trying to match fragments of a voice sample to a word, but rather a more complex system of analysis and "interpretation" indicative of LLMs. Bottom line though is that it's an AI that is generating text. Without a decisive way of determining that a tool addresses the concerns of this policy, I don't think it's possible to allow any of the text generated by those tools.
As an alternative, people could presumably upload their voice recording somewhere and link to that for people to listen to. It's not ideal considering this is a text-based site, but if someone is physically incapable of writing text themselves, that'd be an option. Human transcription is also an option; they could ask a friend or someone in this community to transcribe their voice to text.
i don't understand what is going on, why would we be opposed to using AI or LLM or whatever it is to make TASing easier? why wouldn't Igor's work be acceptable for using AI? If ChatGPT could help me TAS something, i would 100% use it to make it faster or even reach results i can't with my current skillset or knowledge.
Did you read the proposed policy, Igor's submission thread, or any of the other links from Samsara's post before asking these questions?
Have you considered the very real costs and consequences of using this technology before declaring you'd absolutely use it?
Considering you were limited once before, and have shown no improvement or even acknowledgement of your intent or behavior, your access to submissions is being limited again. This time, the duration is indefinite.
I was very clear with you last time. Between this and your other comments it's apparent you still aren't listening to us. Your access to submissions will remain limited indefinitely.
Try using UserFiles instead, as I and many others have suggested numerous times.
In the future, if you've come up with an improvement to a submission that hasn't been judged yet, rather than cancelling and creating a whole new submission we'd prefer you upload the improvement to your userfiles and ask a judge to update the existing submission. That way, forum discussions and the history of a TAS don't become split between different threads.
If you're actively working on an improvement but don't know how long it'll take, we can also delay the submission rather than cancel it.
Huge improvement though! Nice job.
Regarding ChatGPT: ChatGPT and AI in general are tools, just like Lua scripts, savestates, or frame advance. Or do you think I just gave it a prompt and it created the entire TAS gameplay by itself? I don't think so.
I already explained this in the submission, but I'll repeat it: ChatGPT was only used to write code, analyze trace logs, create lua scripts, find addresses and routines, and so on—things I could have done myself, but that would have taken much longer. From my perspective, AI was simply another tool that helped make the development process more efficient.
I don't believe anyone here thinks you just gave it a prompt and AI spit out a complete TAS, but that doesn't mean there aren't implications and costs to using AI. ChatGPT and other LLMs are fundamentally not the same as other TASing tools like scripts of savestates. Are they a tool? Sure. But that's where the similarities end.
ChatGPT doesn't think, nor does it perform deterministic procedures (at least not from a typical user's perspective). It identifies patterns in the character sequences of the prompt, and given a huge set of weights, RNG, and provided context, produces something that would possibly be the natural response to the identified patterns. (What is considered a possible response is based on its training data.) As a result, you'll get text that sounds convincing and likely relates to whatever you've told it previously, but there's absolutely no guarantee or concern that what it says is actually true. Since AI is based on patterns, a lot of what generative AI produces also tends to have distinct patterns. (e.g. an extreme overuse of em-dashes, bullet point lists, and emojis.)
It's fairly clear that at least some, if not all, of the submission text was "written" using generative AI. Which puts into question the accuracy and authenticity of the entire text.
Then there's the costs of AI to consider. Training these AI models takes unfathomably large amounts of power, water, and computing resources. Then even after they are trained, generative AI still consumes a rather large amount of power to actually generate a response, especially if you want the response to be reasonably coherent and produced relatively fast. None of this is anywhere near the costs of writing your own lua scripts, analyzing trace logs yourself or writing a program to do it for you, saving and restoring an emulator state, or advancing a frame in an emulator.
There are many other concerns wrapped up in AI usage too, such as attribution, copyright, and misinformation, as well as promoting an industry that is rapidly destroying communities, small and large businesses, the critical thinking skills of humans, and so much more.
Now, did AI at least partially produce this TAS? In my opinion, yes absolutely.
You may not have used AI to generate the actual frame by frame inputs (or maybe you did, I don't know.) But, there's a lot more to producing a TAS than just throwing inputs into TAStudio. Much of what you described using AI for, were clearly integral to producing this submission. You said it yourself: AI was used to help you to learn, discover, and analyze data. How much more work would AI have to do before you consider it an author of the TAS?
KotyaTAS,
It is clear that you have not heeded the advice/warning you were given the last time your access was limited, nor any of the advice given to you by community members on how to improve your submissions. Your limited access was also re-explained two weeks later.
Please understand that the submission process on this site requires, among other things, that the TAS is reasonably optimized. Frankly speaking, your submissions consistently show little to no effort towards optimization. When feedback is given in the hopes that it'll help you improve your work, you seem to acknowledge its existence by making a reply, but your replies don't address the feedback in any way.
Considering you were limited once before, and have shown no improvement or even acknowledgement of your intent or behavior, your access to submissions is being limited again. This time, the duration is indefinite. You will still have access to the rest of the site, including forums and userfiles, as well as our Discord server. Take time to utilize the resources of this site and its community to improve upon your TASing skills.
Note: If you try to circumvent this restriction, such as by making alt accounts, all of your accounts will be banned. Further misuse of our site or Discord may also result in bans.
Please keep this in mind when posting to the forum, as you've previously made off-topic posts in the wrong threads.
Sounds good. Ban until July 1st, 2028 has been put into effect according to the terms given by Samsara.
Locking this topic as well: no further discussion is necessary here.
Use this thread for official bug reports if possible.
Also, I ask all bugs be reported on our github repository. Though you are welcome (and encouraged) to post them here as well.
TAStudio bugs are reported here.
i downloaded the latest version of the emulator BizHawk 2.11.1
i still cannot get past the first menu of the game
arrow keys up or down it SCROLLS to the TOP of the SCEEEN
is it a BUG?
You already asked this in twoother threads. Don't spam please.
After completing Super Mario Bros. once, you have the option to choose a world, but not a stage within that world. The game state on starting a game this way is always the same regardless of how the game was completed (... mostly).
What isn't the same? Could any part leading up to starting a world have any affect on the actual run? e.g. RNG (idk if that's a thing in SMB, but it could be in other games.)
DigitalDuck wrote:
Should ILs for Super Mario Bros. be worlds selected from the title screen, so each movie can start from a savestate (full game run verified) + soft reset allowing for traditional TAS timing as well as RTA and IGT? I think most people would consider an "individual level" to be a stage within the world, but then time could be affected by whether you start as small Mario, big Mario, or fire flower Mario, and also (depending on the timing method) what score you have at the start, on top of not having a clean "time from power on" or "time from soft reset" TAS time.
The starting state of the game, whether it be at the start of a world or a stage, definitely matters. I think a verification movie is a great approach for handling these questions. It not only does the regular job of verifying the lead up to the actual run is legit, it also sets the stage for what advantages/disadvantages the run receives at the start. For the sake of comparing which run is the "best", one would need to decide what's fair to compare in the first place. As in, is starting a stage with the fire flower just a speed saving technique, or is it a different category of run altogether from runs that start as small mario.
I think this is one of the major issues of this topic to be discussed. Personally, I think the lead up to the start of an IL matters and should be cataloged in some form. As such, it'd make sense to me if there was a different category for each kind of start, depending on what makes sense for the game. Still, there might be minute differences between verification movies for the same category that might lead to real differences in the actual run. I'm not sure how best to handle that. I don't want to say that everyone should be forced to use an prearranged verification movie; that would have its own set of issues and would likely stifle improvements. Maybe we just accept that verification movies can offer advantages and they become a more integral part of a TAS submission.
As for starting from reset or level-start, if we're using a verification movie, I don't think there's any need to start from reset. Alternative timing methods are something we're already working on implementing so this is less of a concern for IL discussions.
DigitalDuck wrote:
do they count as two separate levels for the IL leaderboard?
If an IL leaderboard becomes a thing, I don't think it'll be quite the same as what RTA runners may be used to, and that's okay. Fundamentally, TASing isn't always strictly comparable to human play. How a TAS is compared to another TAS may realistically be different to how an RTA run is compared to another RTA run.
Variations in the game's state may just be accepted as what it is to play a game as a human, since they are often not controllable. However, TASing offers a much much higher degree of control and precision. As such there are many more variables to consider when trying to compare one TAS to another TAS. If a run is trivialized down to just how fast the objective is completed, then it's simple to compare runs against each other. But if we care about how the objective is completed, then the comparison is harder to perform. It may not be something that can be simply displayed as a list of times.
also it's for actual age verification, like having to scan an ID
(1) Require any user of the operating system to provide the
date of birth of the user in order to--
(A) set up an account on the operating system; and
(B) use the operating system.
Seems it only requires the user to provide their date of birth. This would be massively more acceptable than sites/apps using private companies that really do scan our faces or IDs. However, it looks like we'd have to wait for the Commission to actually come up with rules on how this should be performed before we'd know what's expected.
It's also not clear how enforceable this would actually be for free OSes.
