yayr

If anybody with enough motivation can modify any existing harmless image to have the same neural hash as a "tracked database" image this will create too many false positives. Too many false positives make the algorithm useless.

If someone with even more motivation and the means to put those images onto your device via social engineering, exploits or maybe even features and you become the target of a criminal investigation in any jurisdiction you just happen to be at the moment, this makes the algorithm harmful.

How long did it take now to make the Apple algorithm ultimately useless or even harmful?

Apple announcement of neural hashing: 5.8.2021.

Generic algorithm to generate a different matching image: 8.8.2021.

one script was already released 10 days ago here https://gist.github.com/unrealwill/c480371c3a4bf3abb29856c29...

you should read again - systems for redistribution of wealth can be good if they are not broken. Income taxes e.g. are imo a great way to do this, if they cannot be avoided. However, to tax an increase in share price without any transaction is often counterproductive.

agreed. However, what alternatives to preference satisfaction do we have to steer an economy? Is the issue here not the miseducation and misculturization of harmful preferences?

but they apparently still have built a useful organisation. So why take away their power over their organisation if the share price increases?

This may be unpopular, but the hard thing in this discussion is to differentiate between consumption and investment. It does not help much to take wealth in terms of equity paper money away from people who know how to invest it efficiently and distribute it. On the other hand, there are minimum basic needs for consumption of food, health services, shelter, education etc. I agree that the system to finance those services is broken. But then fix the system and don't only focus on punishing people when they build economically viable organisations that even help to support the economy in such a crisis.

One problem is, the CSAM database he talks about is from the US and not managed by Apple. What if another country requires to use a different database and requires to put in other in this country "legally compromising" images? Who audits this? Apple in the respective country? Who will even get notified about this?

see some analysis of the scope of this issue here: https://docs.github.com/en/github/copilot/research-recitatio...

especially: Conclusion and Next Steps.

This investigation demonstrates that GitHub Copilot can quote a body of code verbatim, but that it rarely does so, and when it does, it mostly quotes code that everybody quotes, and mostly at the beginning of a file, as if to break the ice.

But there’s still one big difference between GitHub Copilot reciting code and me reciting a poem: I know when I’m quoting. I would also like to know when Copilot is echoing existing code rather than coming up with its own ideas. That way, I’m able to look up background information about that code, and to include credit where credit is due.

The answer is obvious: sharing the prefiltering solution we used in this analysis to detect overlap with the training set. When a suggestion contains snippets copied from the training set, the UI should simply tell you where it’s quoted from. You can then either include proper attribution or decide against using that code altogether.

This duplication search is not yet integrated into the technical preview, but we plan to do so. And we will both continue to work on decreasing rates of recitation, and on making its detection more precise.

To me it seems, the whole subject requires additional consideration in licensing. It is a little like applying telephone based law to the internet. It will not 100% fit.

If the creators interests are not clearly expressed anymore with a license, we need updates to the license texts.

Let's look at MIT:

____________________

"Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. [...]

____________________

From the license text alone, it would not be clear to me, why anyone could claim that the OpenAI codex or the Github Copilot would require attribution to any of the used MIT source code to generate the AI model. The AI model is simply not a copy of the source or of a portion thereof. It is essentially a mathematical / statistical analysis of it.

Now what about any generated new source? How similar does it need to be to any source to be a copy? At what size of the generated code it qualifies to be a copy instead of a snippet of industry best practice?

Where does the responsibility for attribution lie? Should we treat the AI code generation models like a copy & paste program? Usually you cannot really say where the copy came from 100% - how do you know what factors influenced it?

function randomIP() {

  return int2ip(Math.random()\*4294967296) ;  

}

says it all - better don't "mess" with what you encounter

as in any other space it is hard for most artists to get attention for the work that they doing. If you want to get attention you have to share some of you work for free. However, the images you see on those platforms are usually not high resolution and don't come with any usage rights. Those are then the things artists get paid for hopefully.

why just 1.8B?

It would be awesome, if they could integrate with an existing AVB network, i.e. just use one network port of the device as AVB port

There will be a time, where we have legislation to ban energy-inefficient crypto-currencies from being traded on regulated exchanges. This would make any proof of work currency essentially useless in the legislated areas. My bet is, EU will start, others will follow. This will lead to pressure to convert BTC, ETH etc. finally to proof of stake variants.

Am I right to assume that also parts of the interaction UX are now rendered by canvas? I looked at the source and even text selections etc. do not have DOM tree elements anymore.

I am not sure, that the interface tech with brain invasion is the best route going forward in the space. There is work on non-invasive alternatives, that promise sooner results for some use cases, e.g. https://www.mindportal.com

essentially, some people at Salk Institute for Biological Studies and China’s Kunming University of Science and Technology inserted human pluripotent stem cells into a macaque monkey cell to create an embryo...

further links:

- https://www.statnews.com/2021/04/15/international-team-creat...

- https://www.fiercebiotech.com/research/a-controversial-break...

so what is the objective/cost function here? My hypothesis:

- the longer humanity survives, the bigger the value.

- the more of humanity contributes to that goal, the bigger the value.

- the more of economic activity contributes to that goal, the bigger the value.

and finally, why is this relevant:

The more educated people are aware of this situation and are working in relevant companies, the higher the pressure for the companies to comply towards this objective function.

So, personally, I'd appreciate the phenomenon to move from anekdotal to the direction of 5 sigma.

My main concern would be about reusability. Current battery tech is quite limited with respect to how many charging cycles it supports. So do we need to throw away a bike with structural battery pack then in future, once the charging cycles have expired or someone forgot to regularly keep the battery charged?

while the original title created some resentment, I like the idea of the "5 minute production app". Especially, since it is built on an open platform like gitlab. What I would wish, is to not only take hyperscalers into account. I would like this feature to be available on a basic VPS too.

thanks @dnsmichi for taking responsibility quickly. personally, I think competition is what drives the web tech forward and I appreciate the openness gitlab brings to both the technology and the process you are using. The blog post process you have linked demonstrates this very well. Btw, I dig your flame emoji :-D

this does not seem to be official

the usual figure now is 5 Million vaccinations and 30 reported deaths from blood clotting. Which appears to be not much different from the number without vaccinations.

https://www.nbcnews.com/news/world/oxford-astrazeneca-vaccin...

...and shooting sparrows with cannons becomes shooting mosquitoes with lasers since 2021 ;-) nevertheless, I like the project, much dedication was going into this, lot to learn from...

original article (2020): https://www.nature.com/articles/s41598-020-76301-0

typical DNNs work like function approximators. Obviously they can approximate the trajectory of planets even with a simple architecture. It is certainly an achievement to have those computations successfully performed using discrete fields, which may be more in line with quantum theory. But I don't understand, why that leads to a prove of simulation theory.

The real problem with CNAME-tracking is that it provides the loaded adtech libraries the security privileges from the hosted domain.

This means, tracking libs can now also access everything of the website they are hosted in, attach any event handlers etc. It is essentially the same as hosting the tracking library from the content server, except that the tracking provider can change the source code anytime without requiring any consent.

This is a security nightmare imo.

taking up visual info is much faster, than audio info. Plus you have control over the flow and much more flexible UX.

going into audio mode is a completely different attention pattern, and definitely not casual.

Great tool. You do have your happy user group now, congratulations and cudos for persevering!

For a 202* stack that I would want to use however, I'd like to see more flexible integration with other tech stacks. It would be THE perfect frontend for e.g. NextJS, if it would support React exports and an integration to other state management mechanisms like MOBX to drive dynamic sites.

What about a touch screen? Anything that might have been good about the old touch bar could then be just an app blending in at the bottom / top / side. Aside from obviously needing it once running most iOS apps on the device.

Plus: there is no 16" touch display in the lineup yet, this would be something adding value also to touch apps.

For design work, especially logos I can recommend designcrowd.com

I wonder what the incentive model for the business and sales people selling the ads looks like. I understand that developers may be well intentioned there. However, they build a dual use technology and as long as the incentives for business people managing the system are purely profit oriented it is clear that the system will be used in both ways.