AI & ML data ethics (article link)

Hi BRN folks!

I am both new here and holding only some small stake in BRN, but I am quite deep into the topic of machine learning and artificial intelligence specifically in application for marketing, mostly from the perspective of analysis but also with regards to delivery, ... or combined, ... and in near-real-time, ... ok that's for another post. You can call me an informed contributor...

I have not read a whole lot of the threads here yet (sorry!) but searching through them I did not see on first glance the topic of ethics for ML and AI being discussed. With ethics I mean not only regulations and laws for how to use AI/ML, but also who's doing well in other areas like social application, community building, research and advancing AI/ML etc. pp. To have standards and aligned activities around usage or just to take inventory I think it's an important topic to monitor for BRN and believers as well. Commercially it will at least impact licensees as to how well they can comply with regulators, but also think of the impact BRN could have baking some of that into the underlying technology architecture. For us as human beings it's important so we can help everyone at least to feel better (if I break it down to just that) with the very different approach to technology we will have from here on (think robots).

I have been following these conversations in the industry as part of day2day work for a number of years now, today this article gave a good summary of where I think we are right now. Plus a good view into the industry (and competitors) based on the ontology diagram in the article, alone for scrolling down to that landscape diagram it's worth the click, if you don't read anything just look at that - I am not copy-pasting that now, sorry ;-)

I am not trying to promote this very initiative and I am aware they are many others but the core topic itself is one of the keys for success for all companies dealing with advanced machine learning capabilites IMHO and I thought this might be of interest for the forum here as well, so without further ado: https://data.org/news/a-taxonomy-for-ai-data-for-good/ Oh and btw I didn't see BRN in the diagram ... and there is a link in the article where you can submit to them initiatives they might be missing in the ontology (trueNorth is there for instance). I did not submit BRN, maybe someone can...?

Looking forward to any comments, questions or feedback,
and nice to (virtually) meet here,
Chris
P.S. Just got off the phone with @cosors after 90 minutes on this very topic, he's also the one who recommended this forum to me... for disclosure...
 
  • Like
  • Love
  • Fire
Reactions: 24 users

cosors

👀
Hi BRN folks!

I am both new here and holding only some small stake in BRN, but I am quite deep into the topic of machine learning and artificial intelligence specifically in application for marketing, mostly from the perspective of analysis but also with regards to delivery, ... or combined, ... and in near-real-time, ... ok that's for another post. You can call me an informed contributor...

I have not read a whole lot of the threads here yet (sorry!) but searching through them I did not see on first glance the topic of ethics for ML and AI being discussed. With ethics I mean not only regulations and laws for how to use AI/ML, but also who's doing well in other areas like social application, community building, research and advancing AI/ML etc. pp. To have standards and aligned activities around usage or just to take inventory I think it's an important topic to monitor for BRN and believers as well. Commercially it will at least impact licensees as to how well they can comply with regulators, but also think of the impact BRN could have baking some of that into the underlying technology architecture. For us as human beings it's important so we can help everyone at least to feel better (if I break it down to just that) with the very different approach to technology we will have from here on (think robots).

I have been following these conversations in the industry as part of day2day work for a number of years now, today this article gave a good summary of where I think we are right now. Plus a good view into the industry (and competitors) based on the ontology diagram in the article, alone for scrolling down to that landscape diagram it's worth the click, if you don't read anything just look at that - I am not copy-pasting that now, sorry ;-)

I am not trying to promote this very initiative and I am aware they are many others but the core topic itself is one of the keys for success for all companies dealing with advanced machine learning capabilites IMHO and I thought this might be of interest for the forum here as well, so without further ado: https://data.org/news/a-taxonomy-for-ai-data-for-good/ Oh and btw I didn't see BRN in the diagram ... and there is a link in the article where you can submit to them initiatives they might be missing in the ontology (trueNorth is there for instance). I did not submit BRN, maybe someone can...?

Looking forward to any comments, questions or feedback,
and nice to (virtually) meet here,
Chris
P.S. Just got off the phone with @cosors after 90 minutes on this very topic, he's also the one who recommended this forum to me... for disclosure...
I have just heard in the news for the first time about a decision of the European Court of Justice ECJ which goes roughly in the direction of your looking into the future topic. It's about passenger data.

German article:
"It was not okay for self-learning artificial intelligence to take over, i.e. for the machine to determine ever new human characteristics that are considered suspect. For the first time, the ECJ is limiting the use of artificial intelligence here - for this reason alone, this ruling is significant, because this could be a model for future decisions on completely different issues."

First I could found in Englisch:
"The European Court of Justice ruled on Tuesday (21 June) that the EU’s Passenger Name Record (PNR) directive must be curtailed to be compatible with fundamental rights."
https://www.euractiv.com/section/da...ravel-surveillance-to-the-strictly-necessary/
____
Keep in mind that this is mostly about cloud data but also the decisions AI makes. This is analogous to our topic with autonomous driving.
 
Last edited:
  • Like
  • Fire
  • Thinking
Reactions: 5 users

stuart888

Regular
Howdee! Seems like the ethics is this: How can improved ADAS crash stoppage Not Be implemented. Lives are lost everyday, so the sooner the statistical proof of mileage driven with and without accidents, it would be tough to keep better safety out. Keep adding on the safety features, Blind Spot Monitoring is a great start, Emergency Braking better next. Let's just get level 2 and 3 going, and maybe in 10 years level 4+.

When airbags went in, they were not perfect, but saved a lot of lives. Sure would be nice to have automobiles not crash, and therefore not waste money on airbags at all. Just some quick thoughts.
 
  • Like
  • Love
Reactions: 5 users

cosors

👀
This article summarises several issues that affect us specifically (Tesla facial recognition; several lawsuits are pending in US courts and in Germany) and generally in terms of this thread and data ethics.

"JUSTIN LING - SECURITY - JUL 1, 2022 7:00 AM

Is Your New Car a Threat to National Security?

Putting sensor-packed Chinese cars on Western roads could be a privacy issue. Just ask Tesla.

STARTING THIS WEEK, Teslas won’t be welcome in the Chinese resort town of Beidaihe. The electric cars are strictly banned on the streets of the coastal city for the next two months, as senior Communist leadership descends on the city for a secret conclave.

It’s not the first time, either. The city of Chengdu barred Teslas in advance of a June visit from President Xi Jinping, Reuters reported, while some military sites have similarly forbade Elon Musk’s flagship product. While no official reason was released, the bans seem to be out of concern that the vehicles’ impressive array of sensors and cameras may offer a line of sight into meetings of Beijing’s senior leadership.
It’s a curious move. China is, increasingly, one of the most connected countries in the world—Chinese industry has even tried to brand Chengdu as the “5G Joy City,” where locals are encouraged to stream their daily lives.

Tesla may be one of the most popular electric vehicle brands in China, with upwards of a half-million vehicles on the roads, but it is not itself Chinese. The firm has acquiesced to Beijing’s data localization demands, setting up a dedicated data center in China, but it cannot shake the characterization that it is a foreign company—and, therefore, a national security threat.
It’s not a concern unique to Xi’s government. As Chinese automakers gear up for a big push into the West, anxieties are already mounting as to how those vehicles could phone their robust trove of data home.


The future of transport is certain to be electric and autonomous vehicles. They could also be the future of espionage.

NATIONAL ANXIETY ABOUT the surveillance powers of new modes of transportation is hardly novel.

In 1913, the French army seized the German-made Z-4 airship after it flew off course in thick fog and landed on French soil. Paris ordered that “any photographs of French fortified places taken en route would also be seized,” The New York Times reported at the time.


Through the Cold War, both sides of the Iron Curtain addressed the question of expanding aerial surveillance capabilities by signing the Open Skies Treaty—opting to provide clear rules on how and when both NATO and Warsaw Pact countries would spy on each other from the skies, even regulating the flight path for these surveillance missions, instead of attempting to stop them outright.
Consumer vehicles are just a recent addition to the national security equation. But thanks to the globalized economy and modern product development, they are perhaps the trickiest challenge yet.
As it stands, Teslas are arguably the most connected and widespread of a new generation of vehicles. Not only do they hoover up a massive amount of data on the driver—from call logs to on-board browser history to average speed and route history—but their outward-facing sensors and cameras can relay a considerable amount of information about the surrounding world.

David Colombo, a 19-year-old German programmer, proved earlier this year that accessing incredibly sensitive data on Tesla users wasn’t just possible—it was fairly easy. Using a third-party application with access to Tesla’s API, Colombo got into the systems of more than two dozen Teslas around the world, controlling their locks, windows, and sound systems and downloading a huge bundle of information.
“I was able to see a large amount of data. Including where the Tesla has been, where it charged, current location, where it usually parks, when it was driving, the speed of the trips, the navigation requests, history of software updates, even a history of weather around the Tesla and just so much more,” Colombo wrote in a Medium post published in January that detailed his exploits.
While the specific vulnerabilities Colombo took advantage of have been patched, his hack demonstrates a huge flaw at the core of these smart vehicles: Sharing data is not a bug; it’s a feature.
The amount of data Tesla collects and uses is just the tip of the iceberg. We have yet to see fully autonomous vehicles or the much-vaunted “smart cities,” which could see 5G-enabled roads and traffic lights.

In the near future, cars will not only collect information about their driver and passengers, but the vehicles, pedestrians, and city around them. Some of that data will be necessary for the car to function properly—to reduce collisions, better plan routes, and improve the vehicles themselves.
“The United States and Europe have been asleep at the wheel,” says Tu Le, managing director of Sino Auto Insights. The US, Canada, and Europe may continue to be the world leaders in producing traditional vehicles, but that lead won’t hold for long. Whether it’s cobalt mining, lithium battery innovation, 5G-enabled technology, or large data analytics, Le says China has been several steps ahead of its Western competitors.
“All those seemingly unrelated things are converging into this smart EV,” Le says.
Of course, not all of Beijing’s success came honestly. Chinese nationals have been accused of pilfering intellectual property from American companies to bolster China’s growing industry. Le says that sort of espionage certainly helps, but it’s not the main reason for Beijing’s exploding growth in the automotive sector.



The West, meanwhile, has been sluggish in adopting local data and privacy protections.
As it stands, Le says, Chinese electric vehicles are roughly three years away from hitting American streets in a major way. “They’re already in our backyard, and we haven’t done anything yet,” he argues.

It’s not just about regulating Chinese vehicles once they arrive, either. As Colombo’s hack showed, domestic vehicle manufacturers need to step up their security game as well. Many manufacturers push software and firmware updates for various aspects of their vehicles over the air.
“Think about the danger when an update is sent to hundreds of thousands of cars wirelessly,” wrote Alexander Poizner, CEO of UK-based cybersecurity firm Parabellyx, in a 2021 blog post. He posed a hypothetical: “What if China used malware to disrupt traffic in Taiwan as a prelude to a military attack?”
Insufficient regulation has led to a total lack of consistency, as Poizner noted: “There is no single standard around cybersecurity for either autonomous vehicles or the infrastructure to support these across the automotive industry.” But cybersecurity standards aren’t the only area where the US is coming up short.
“Policymakers are struggling at the highest level,” says Marjory Blumenthal, senior fellow and the director of the Technology and International Affairs Program at the Carnegie Endowment for International Peace, a global think tank based in Washington, DC.
Nevertheless, Washington’s instincts may be quite similar to Beijing’s. In the past, the United States and its allies have opted to simply ban Chinese products from sensitive areas—from the country’s unsuccessful TikTok ban to its considerably more effective effort to exclude Huawei technology from 5G infrastructure. America’s allies have followed suit in blocking Huawei from the backbone of their next-generation mobile systems, including Australia, Canada, and the United Kingdom.
In 2018, the Trump administration moved to slap tariffs on the Chinese automotive sector, arguing that the foreign competition threatened to undermine America’s domestic industry, thus harming a research-and-development pipeline into the US military. “It is imperative that related R&D remain within the United States, be conducted by American-owned firms, and that the United States Government take measures to secure the long-term viability of domestic R&D in the automotive sector,” reads a 2019 Commerce Department report. (The tariffs were later abandoned.)
Such a protectionist move would likely kneecap major Western automakers, which are currently vying for new market share in China. Beijing has made it clear that any protectionism in the West would be met with retaliatory measures.
There are certainly concerns that curtailing how vehicle data can be collected, analyzed, and transferred could limit research and development of automotive companies looking to keep up with their Chinese competitors, Blumenthal says. Canada and the European Union do have more expansive and consistent privacy laws that offer a clearer road map for companies headquartered there, unlike the United States. “The data questions are less well explored in this country, given that we don’t have a monolithic privacy regime,” Blumenthal adds.


As companies hustle to build out these new systems, Blumenthal says, they will be collecting a huge volume of information. “That then raises the question of how much is stored? Where is it stored? For how long is it stored?” she says. Governments need to regulate these areas, she adds, and worry less about China’s panopticon model.

There may be grand claims about what China hopes to do with its unparalleled heap of data, but Blumenthal says she’s not convinced that China’s system will be better simply because it captures more data. “I’m not ready to buy that.”
As the technology matures, she says, companies may figure out how to reduce the noise in that data, collecting only what is necessary to improve safety, make routes more efficient, and inform innovation.
Creativity in determining how those algorithms work may ultimately mean more than the data feeding into it, she says.
Le says there’s a desperate need for clarity—rules about what data can be freely exploited, what data needs to be anonymized, and what needs to be held within a country’s borders. “We’re over-relying on the tech industry to say, ‘Oh, we’ll keep it safe for you,’” he says.
“We might look back in 10 years and see it’s the frog-boiling scenario,” Blumenthal says of the auto industry’s increasingly sophisticated data collection. Or, she adds, “we’ll have a scenario where people are adapting to all the behavioral monitoring in the world.”
But there’s a note of optimism. While legislative fixes to address vehicle data collection have wallowed in Congress, Blumenthal points to the National Highway Traffic Safety Administration’s efforts to modernize its policies to keep up with the times. “As they do that, it might be reasonable to assume that they could add privacy there,” she says.
China may be a walled garden for this technology, but the West has a history of determining the rules of the road collectively. “There is a framework of international standard-setting—and in the last two to three years you’ve seen an increase in standard-setting,” Blumenthal says.
How the world handles the data at the heart of these smart vehicles will ultimately determine the urgency of security concerns. Clear, consistent rules across the major economies could allay espionage fears and decrease the likelihood that competitors will set out to hack each others’ vehicles. Strong encryption, privacy protections, and other data regulations could help prevent the weaponization of drivers’ personal vehicles.

With the right constraints in place, the data collected by these vehicles could limit espionage and national security threats while significantly reducing crash fatalities and speeding up research and development.
Cooperation with Beijing could accelerate that process. Bitter competition could slow it all down."
https://www.wired.com/story/china-cars-surveillance-national-security/
 
Last edited:
  • Wow
  • Like
  • Fire
Reactions: 7 users

uiux

Regular
Professor Katina Michael is always good for a read:



She was a guest on the BrainChip podcast:

 
  • Like
  • Fire
Reactions: 6 users

uiux

Regular
  • Like
  • Fire
Reactions: 3 users

cosors

👀
Nviso professes to:

"ETHICAL AND TRUSTWORTHY
AI systems need to be resilient and secure. They need to be safe, ensuring a fall back plan in case something goes wrong, as well as being accurate, reliable and reproducible. Additionally unfair bias must be avoided, as it could could have multiple negative implications. NVISO adopts Trustworthy AI frameworks and state-of-the-art policies and practices to ensure its AI Apps are "fit-for-purpose".

HUMAN BEHAVIOUR
AI SDK
 
  • Fire
  • Like
Reactions: 5 users

cosors

👀
Here is a thought experiment. Read carefully!

"The end of irrelevant artificial intelligence
...
An interesting anecdote about the use of AI chatbots is the story of Xiaoice, a chatbot developed by Microsoft in China. Xiaoice was designed to have natural, human-like conversations with people, and she quickly became popular with users who enjoyed talking to her. Many users took Xiaoice so much to their hearts that they didn't even realise she was a chatbot, and some even claimed to be in love with her.

In the near future, AI chatbots will be an integral part of our daily lives. These intelligent, conversational agents will be able to assist us with a variety of tasks, from the mundane to the complex. The future of chatbots, and ChatGPT in particular, is likely to be one where they are an integral part of our daily lives. ChatGPT, a large language model trained by OpenAI, has already demonstrated the ability to have natural, human-like conversations across a wide range of topics. This capability, combined with the convenience and accessibility of chatbots, makes them a promising technology for everyday use.

One of the most interesting use cases for AI chatbots is customer service. Chatbots can handle a large number of customer queries, allowing human customer service agents to focus on more complex issues. Chatbots can also provide quick and accurate answers to common questions, improving the customer experience. Another interesting use case for AI chatbots is healthcare. Chatbots can be used to provide patients with information and help them manage their health and make informed decisions. For example, a chatbot could provide information about symptoms and treatment options or remind patients to take their medication. This can be particularly useful for people with chronic conditions who need ongoing support.

A third interesting use case for AI chatbots is education. Chatbots can be used to provide students with personalised learning experiences, tailoring the content and pace of lessons to each individual's needs. For example, a chatbot could help a student study for an exam by providing practice questions and feedback. This could be a valuable tool to help students learn more effectively. They could be used in finance, providing personalised investment advice and helping people manage their finances more effectively. And they could even be used in entertainment to provide engaging and personalised experiences for users.

However, there are also potential dangers associated with the widespread use of ChatGPT and other chatbots. One concern is the potential for abuse and manipulation. Chatbots, like any technology, can be used for nefarious purposes. For example, they could be used to spread false information or to harass and intimidate others. This is particularly worrying because chatbots are capable of having persuasive conversations with people. Another potential danger is the possibility of chatbots replacing human interaction. Although ChatGPT and other chatbots can provide valuable help and convenience, they should not be seen as a substitute for human contact. In some cases, people might rely too much on chatbots and lose their ability to communicate effectively with others. This could have negative consequences for both individuals and society as a whole.

Furthermore, the use of ChatGPT and other chatbots raises ethical questions. As these technologies become more advanced, they may be able to perform tasks that were previously only possible for humans. This could lead to employment issues and displacement of workers. It is important for society to consider and address these ethical concerns as the use of chatbots becomes more widespread.

Overall, the future use of ChatGPT and other chatbots has the potential to significantly improve our daily lives. However, it is important to be aware of the potential dangers and take measures to mitigate them. This may include careful regulation of the use of chatbots and ongoing dialogue about their ethical implications. By addressing these issues, we can ensure that ChatGPT and other chatbots are used in a responsible and useful way."

________
________

Sascha Lobo: "I didn't write a single word of the italics above in this column. It comes 100 percent from an artificial intelligence called ChatGPT and that, to quote a popular German chancellor, is a turning point.
...

However, because this has only worked in English so far, I had the above column translated into German by the currently best AI translation service Deepl.com "

https://www.spiegel.de/netzwelt/web...olumne-a-b2afeb69-083d-4e69-8920-da5cad549d5f

That's interesting, isn't it?
 
Last edited:
  • Like
  • Thinking
Reactions: 3 users
Top Bottom