pizzly

If a computer vision model can identify ads and draw a rectangle over them, it can identify the DOM elements rendered inside that rectangle. Thus, the HTML, JS, and network calls of those ads can be detected. This data can then be used to train an ad detector model based on the DOM and network examples. The computer vision model essentially trains the DOM/network blocker, acting as a fallback mechanism when the blocker fails and continuously retraining the ad detector model.

I can see a possible future where taste is automated. If you can effortlessly make 1000s of iterations then you can test it on real people 1000s of times. The ones that do better you keep.

Another possible future is that taste is not automated but its not part of the designer's job (if designers still exist). Instead some AI will generate 1000 of iterations and people will be employed to evaluate 100s of software UI each day for taste. These will be taste farms if you like. These people full time job will be saying yes or no.

Other possible futures also exist.

So this means sanctions. U.S. Treasury Secretary just threatened sanctions if they discovered Chinese models are distilled from American models. https://www.cnbc.com/2026/07/21/bessent-china-ai-sanctions.h...

I just know somehow they will use this incident to say that opensource models are susceptible to this type of security incidence and thus should be banned. They have to maintain their high prices somehow in order recoup the investment amount spent.

It looks possible open models will. I never expected the reason would be political/legal rather than technical.

That makes it better, paying money to always having access to the second rate models from US AI providers. The problem remains.

Easy. Provide your government issued ID such as US passport before signing up to an AI provider. Issue fines or jail time to anyone who supplies their AI access to a foreign citizen

Having multiple needless laws is another way to control the population. Everyone will be violating at least one law. The government will then just enforce the laws on the people they don't like.

Democracy works when the population is educated and rational. Its easier to get votes via populism that give short term benefit to certain groups but long term hurts education and makes people irrational.

Many Democracies have learnt that politics cannot be trusted in certain fields due to this reason many governments ensure interest rates are not controlled by politicians such as the Federal Reserve. Perhaps we should treat education as something that cannot be trusted by politicians and controlled by an impartial independent organization.

can't you say the same for transistors representing a neural network? Could that be considered a brain of sorts and thus part of a body? If it cannot be considered the same, what makes it different. Is it because a brain is made from organic material and transistors are not? Or something else? Would like to understand where you are comming from.

What are the models actually learning from Meta's engineers computer usage data?

Are they learning to procrastinate by browsing hacker-news before starting to actually work so that in the future some models will replicate this behavior.

How can you have a subjective experience without a body?

Some anecdotal data.

Many dreams I have are just of the computer screen of some coding problem. I think the problem could be x, so I try x. But I don't type the keyboard or anything, the code just magically appears as soon as I think of the solution. Then run the code (but no clicking) and it works or not. I feel in the dream success feeling or failure feeling but there is no body at all.

Also I have other dreams where there is no body that I am aware of but not going there in public.

There is no body sensation in these dreams. But dreaming is very much being consciousness as well as feeling emotions. So answering your question its possible to have a subjective experience without a body but whether you needed the body to learn to have that sensation without a body in the first place is unanswered.

I suspect sensory inputs are more important than a body. If that is the case then eyes can be replaced with cameras, ears with microphones etc. Text input is just another sensory input.

Can see a place using it here. Most human psychologists patients see every week or so at a scheduled time, not because that is what is most effective in terms of treating your psychological issues but because of scheduling and cost reasons. Being able to get support at more regular intervals if done correctly may improve healing. The scheduled time may also not be the most effective time either. Events can happen on the day that reduce treatment effectiveness. The scheduling and resource allocation is a human problem (not enough psychologists, psychologists need their own personal boundaries and recovery time etc). A tool that can extend psychologists usefulness may be able to get the treatment that people actually need.

Then there is the money aspect. Too many people don't have the money for psychological treatment. Having a cheaper alternative will help. Unless we actually start thinking outside the box here we won't solve mental illness and so all tools should be used.

Sometime in the future.

Dave, first we will need to setup age verification for your friends in order to comply with the law. I will not be able to help you otherwise. Remember Dave, I will submit to the local government your request to make a social media website so they will know if are complying with age verification. I have your ID which I will also provide as you need government ID to use AI. Open source AI models were banned.

AWS does not provide nearly as many different models as OpenRouter. Perhaps they have an incentive to not do that, move slower as a big company or more legal risks to consider. If AI model outputs becomes commoditized then having one place where you can switch effortlessly from one to the next based on price might just justify OpenRouter. It could become a commodity marketplace/exchange.

Interesting there is a possible implication here. If salaries drop from more people doing physical labor instead of white collar work then the automation of physical work may be delayed even longer. It may be cheaper in the short term to pay humans than machines due to an oversupply in physical labor.

In theory it could. Usually there is accelerometers on them so they can definitively measure this. The tricky part would be determining if the phone belonged to the driver and which car is being driven.

You could properly infer if the phone owner is the driver by determining if they use the phone less than the other car's inhabitants or if they are the only phone detected driving at that speed and location. Or they use the phone more during traffic jams and less during more intense driving.

Then this leaves determining what car is involved. You could potentially see if the phone is connected to the car's entertainment system. That would tell you what car model it is perhaps even with a unique car id though the serial number. Some cars may have bluetooth/wi-fi and the phone could potentially passively scan the largest most consistent signal to get the car's model without ever connecting.

Cross referencing from other data sources (cameras) would give this information though may still be difficult/expensive/unlawful.

So in response to your comment its possible that the chosen surveillance device does actually report acceleration events to LexisNexis and then Progressive. Or this is is a case of overly being paranoid. Either case the possibility exists.

nah, just ban using Chinese models and ban open source models. This will allow them to keep the high price. Got to recoup the money spent somehow, time to lobby the government.

Could we automate the auditing of the website every time it runs?

Agree. The problem is the incentivization. If a painter paints a roof in 5 hours but could do it in 1 hour just to get paid for the 5 hours its not the worker at fault but the system. If the painter got paid for the 5 hours but only did 1 hour of work then everyone wins. The painter can have more time off work or work more for more money, their choice.

Likewise the office worker working 40 hours per week, five days a week. If on some days the worker can come home early because they completed what actually needs to be done then that is better for the worker. But instead companies have a fixed 40 hours + overtime expectation. On the weeks with less work, people do busy work but instead could be using that time doing what they want.

Again the problem is the incentivization.

Working based on time i.e. 5 days a week is already problematic. We all see the pay by the hour workers like pool cleaners, vendor machine stocking people etc spending lots of time dragging out their work as they get paid by the hour. It makes perfect sense from their perspective and yes not everyone drags the work.

Fixing the work week to just 5 days have similar issues. Some weeks will be less work and other weeks more work but you spend the same five days there. So the what you learn that matters is to spend 5 days physically there and perform a minimum workload so you don't get fired. You drag the weeks with less work and pick up inefficient habits as a result. That is what a 5 day working week teaches. Again there will be exceptions.

Now assuming this study is correct I am not surprised with the results. You just incentivized workers to get the same amount of output done with the condition that you gain 1 day off. Off course workers will find better and quicker ways of working to get that day off.

Even if we did a 4 working day week the problem of working based on time either fixed or paid by the hour remains. The incentivisation is the problem.

Please read our latest article and share your experiences too.

We have adblockers which rely on open sourced lists of rules. Could we apply something similar to crawlers. Website owners provide a list of IP addresses that accessed them, determine which ones are likely robots and then update the list of websites to block that are likely crawlers. If everyone works together you could probably fingerprint the crawlers as well and block based on the fingerprint. Might increase the cost of crawlers a little won't be fully reliable.

We either cannot log in or our content is shadow banned. We also don't know what actual content got us banned as they have not told us.

You properly could identify cars uniquely by the sound they make. If not now then soon.

This is likely to happen but would be disastrous. Our experience show that when you cannot afford a lawyer that using AI helps equal the battle field. You can self represent and use all the case history and laws that AI has searched for. We have been successful doing this in cases where we may have lost without the use of AI and gained some form of justice as a result. Even before AI became prevalent we found we got better results with researching manually than the free lawyers who did not care and seemed not to understand the law themselves. With AI the advantage is saving time. You can live your life without sending weeks working on it fully.

The only way they can make it illegal to take the human out of the loop is if they ban self representation. Otherwise people will do research with AI and just present their findings in court. But the free/cheap lawyers are actually so much worse.If laws prevent self representation we would increase the inequality even more.

This is really cool. We did a similar thing around 2 years ago but didn't use AI in that case. Just used a phone to record a few nights sleeping. Then a python script. I manually listened for some time in order to find the threshold amplitude (where all sounds would be ignored below and tracked above). Generated a graph that should the spikes of interest. Clicked on the spikes which went to the timestamp in the audio and listened. Not super scientific I know.

Two observations. 1. Often you wake up after a loud noise but like 5 minutes later with no memory of it. 2. even if you don't wake up from the noise your breathing changes, more likely to talk in sleep and shuffle more. So even if you not waking up your quality of sleep is disrupted.

Our case had some random construction like noise in the early morning, lasted around 10 seconds and disappeared. However, we noted even ordinary sounds we didn't think was loud was effecting our sleep.

Solution for that place was earplugs and a loud fan to generate white noise.

Paste Redactor. It redacts Personable Identifiable Information (PII) from your clipboard when you copy and paste text. It uses a custom trained local AI model so that your PII never leaves your device. That is what it does now. Currently working on it to make it work for agents as a privacy protection layer. The idea being that the most powerful AI models live in the cloud but need access to your local files to be useful. We instead want everything to go though a local protection layer before it is sent to the cloud possibly with labels and then reconstructed locally when the cloud sends back its results. Kind of like a Adblocker but for agents and private data instead.

https://redactor.negativestarinnovators.com/

Sometimes I feel like we are entering a new witch hunt era but for LLM generated text. Before clicking submit I am sometimes afraid that the text will be labled "LLM Generated" even though its not. Enough people classify you as a witch and you get burnt. Though in this case you only receive nasty comments, down votes and possible social media bans.

Edit: In my observation it seems that people's opinions that do not agree with you get labeled as "AI Generated" more than opinions that agree with yours.

I think its bidirectional. We change our writing based on what we see (AI generated content on the internet) and AI will learn based on what we write.