Free Resource on AI for Physical Simulations
September 27, 2021
The academics at the Thuerey Group have made a useful book on artificial intelligence operations and smart software applications available online. The Physics-Based Deep Learning Book is a comprehensive yet practical introduction to machine learning for physical simulations. Included are code examples presented via Jupyter notebooks. The book’s introduction includes this passage:
“People who are unfamiliar with DL methods often associate neural networks with black boxes, and see the training processes as something that is beyond the grasp of human understanding. However, these viewpoints typically stem from relying on hearsay and not dealing with the topic enough. Rather, the situation is a very common one in science: we are facing a new class of methods, and ‘all the gritty details’ are not yet fully worked out. However, this is pretty common for scientific advances. … Thus, it is important to be aware of the fact that – in a way – there is nothing magical or otherworldly to deep learning methods. They’re simply another set of numerical tools. That being said, they’re clearly fairly new, and right now definitely the most powerful set of tools we have for non-linear problems. Just because all the details aren’t fully worked out and nicely written up, that shouldn’t stop us from including these powerful methods in our numerical toolbox.”
This virtual tome would be a good place to start doing just that. Interested readers may want to begin studying it right away or bookmark it for later. Also see the Thuerey Group’s other publications for more information on numerical methods for deep-learning physics simulations.
Cynthia Murrell, September 27, 2021
US Government Procurement: A Technology Brake?
September 27, 2021
I read “Study: Pentagon Reliance on Contractors Hurt US in 9/11 Wars.” I was not certain how to process the story. Was it a blockbuster exposé or was it another recycled Hummer tire?
The write up states:
Up to half of the $14 trillion spent by the Pentagon since 9/11 went to for-profit defense contractors, a study released Monday found. It’s the latest work to argue the U.S. reliance on private corporations for war-zone duties that used to be done by troops contributed to mission failure in Afghanistan. In the post-9/11 wars, U.S. corporations contracted by the Defense Department not only handled war-zone logistics like running fuel convoys and staffing chow lines but performed mission-crucial work like training and equipping Afghan security forces — security forces that collapsed last month as the Taliban swept the country.
Has the enshrinement of procurement methodology created the situation? Are there other forces at work; for example, people complain about meetings. Nevertheless, the work of some government professionals is meetings.
Who does the work?
Maybe contractors? Interns? People hired on Fiverr?
The write up states:
And up to a third of the Pentagon contracts went to just five weapons suppliers. Last fiscal year, for example, the money Lockheed Martin alone got from Pentagon contracts was one and a half times the entire budgets of the State Department and the U.S. Agency for International Development, according to the study.
Are the expenditures audited? Does anyone know where the money goes?
The write up wraps up with this statement:
Relying less on private contractors, and more on the U.S. military as in past wars, might have given the U.S. better chances of victory in Afghanistan…
Interesting.
Stephen E Arnold, September 27, 2021
Telegram and Criminal Usage: Who Knew?
September 27, 2021
Why would cyber criminals and regular run-of-the-mill criminals use a message app which was able to encrypt messages, enable “transactions,” and support file attachments? (A file attachment could be malware, an image one would not show a grade school class, or a video with semi-interesting behavior on display.)
“Telegram Has Seen a Sharp Rise in Cybercriminal Activities, Report Says” reveals this previously unknown factoid. Astounding. I learned:
Cybercriminals have been using Telegram for years, because it’s encrypted and easy to access. According to a recent investigation conducted by The Financial Times and cyber intelligence group Cyberint, though, there’s been “a 100 percent-plus rise in Telegram usage by cybercriminals” recently. And FT says the rise in criminal activity on the app came after users flocked to it following a change in WhatsApp’s privacy policy.
Yep, Facebook — again.
The write up did not make clear that:
- Telegram has reached some rapprochement with Russia’s telecommunications authority.
- Encryption at scale creates interesting challenges for law enforcement, intelligence, and regulatory entities
- Fosters a wide range of criminal activities; for example, recruiting individuals for illegal activities, dissemination of proscribed content, and coordinating distributed cyber crime actions.
Is there a solution? Not an easy one I fear.
Stephen E Arnold, September 27, 2021
Yay, A Facebook Friday
September 24, 2021
Three slightly intriguing factoids about the Zuckbook.
The first is a characterization of Facebook’s and the supreme leader’s time spirit:
“Shame, addiction, and dishonesty.”
Well, that’s a poster message for some innovator in the decorative arts. The original could be offered on Facebook Messenger and the cash transaction handled at night in a fast food joint’s parking lot. What could go wrong? And the source of this information? The work of the UX Collective and included in a write up with the title “Zuckerberg’s Zeitgeist: A Culture of Shame, Addiction, and Dishonesty.” What’s left out of the write up? How many UX Collective professionals have Facebook accounts? And what’s the method of remediation? A better interface. Okay. Deep.
The second is from “Facebook’s Incoming Chief Technology Officer Once Said People Being Cyberbullied to Suicide of Killed in Terror Attacks Organized on the Site Was a Price Worth Paying to Connect People.” The headline alleges that the new Facebook chief technology officer or C3PO robot emitted this statement. Another memorable phrase from the C2PO Facebooker is allegedly:
Maybe it costs a life by exposing someone to bullies. Maybe someone dies in a terrorist attack coordinated on our tools. And still we connect people.’
Snappy? Yep.
And, finally, today (September 24, 2021), that the estimable Salesforce luminary, Marc Benioff, who maybe said:
In regards to Facebook, they are not held accountable.
The write up “Tech Billionaire: Facebook Is What’s Wrong with America” contains an even more T shirtable slogan. I live in fear of Google’s duplication savvy smart software, but I want to be clear:
Facebook is what’s wrong with America
I like this statement whether from the humanoid running Salesforce or a thumbtyping PR expert with a degree in art history and a minor in business communications. Winner.
Net net: Facebook seems to be a font of news and inspiration. And, please, remember the fix: user interface changes. Yes.
Stephen E Arnold, September 24, 2021
Ethics Instruction: Who Knew?
September 24, 2021
Well, this is not particularly alarming. Despite increasing concern over the harm caused by unbridled algorithms, many AI students are still not being taught ethics in their coursework. The Next Web reports, “Data Science Students Don’t Know a Lot About Ethics–and That’s a Problem.” Ethical problem-solving is specifically mentioned in the National Academies recommend 10 training areas for data-science degrees. Considering the dramatic rise in students going into this field, the authors investigated the instruction undergraduates are receiving. They write:
“In our study, we compared undergraduate data science curricula with the expectations for undergraduate data science training put forth by the National Academies of Sciences, Engineering and Medicine. Those expectations include training in ethics. We found most programs dedicated considerable coursework to mathematics, statistics and computer science, but little training in ethical considerations such as privacy and systemic bias. Only 50% of the degree programs we investigated required any coursework in ethics. Why it matters: As with any powerful tool, the responsible application of data science requires training in how to use data science and to understand its impacts. Our results align with prior work that found little attention is paid to ethics in data science degree programs. This suggests that undergraduate data science degree programs may produce a workforce without the training and judgment to apply data science methods responsibly. … We believe explicit training in ethical practices would better prepare a socially responsible data science workforce.”
The study focused on R1 schools, or those with high levels of research activity. The authors note there may be more ethics instruction to be found at schools with lower levels of research or in graduate-level courses. It seems like more research is needed.
Cynthia Murrell, September 24, 2021
Google: More Management of Sensitive Issues
September 24, 2021
Some MBA engineers are driven purely by greed without regard for their fellow humans. When Google formed its parent company, Alphabet Inc., they changed their company motto from “Don’t be evil” to “Do the right thing.” However, Google has proven it does not do the right thing when it comes to respecting user privacy and pursuing the almighty dollar. Google has violated user privacy multiple ways, while they tried to establish a market in China despite the country’s abhorrent human rights record.
The Daily Hunt explains that, “Alphabet Inc’s Google Gave User Data To Hong Kong Authorities Despite Vow.” The Hong Kong Free Press reported that Google gave the Hong Kong government user data, despite promising not to do so. Google said that these reports were actually stop bad actors and crime:
“Alphabet Inc’s Google complied with three of 43 government requests received between July and December 2019, the company told HKFP. One request was for an emergency disclosure involving a credible threat to life, Google said, while the others involved human trafficking and were supported by search warrants granted by the court. They were not related to national security and no user content data was shared, the company added.”
Other technology companies, including Microsoft, Facebook, and Twitter, said they would no longer comply with Hong Kong government data requests, because China imposed a national security law that violates civil rights.
Google could be telling the truth when it comes to preventing human trafficking and saving human lives, but they could also comply with the Chinese government in order to gain favor in its technology market.
Huge corporations pretend to be ethical, but its usually lip service. Money and the threat of bad publicity has more sway than violating civil liberties and human rights. Google is not any different.
Whitney Grace, September 24, 2021
NSO Group and Collateral Damage: Shadowdragon
September 23, 2021
The NSO Group has captured headlines and given a number of journalists a new beat to cover: Special service vendors. This phrase “specialized service vendors” is the one I use to capture the market niche served by companies as diverse as Anduril to Voyager Labs. Most of these firms walk a fine line: Providing enough public information so that a would-be customer like a government agency can locate a contact point to staying out of the floodlights looking for next NSO Group to research and write about.
I read “Shadowdragon: Inside the Social Media Surveillance Software That Can Watch Your Every Move.” The exposé appearing in the The Intercept follows a predictable pattern: Surveillance, law enforcement, technology, sources, similar software (in this story Kaseware), and rights violations.
A Wall Street Journal reporter is allegedly working on a book that will surf on the the NSO Group’s tsunami of surveillance shock.
I have spelled out three concerns about what I call the conversion of NSO Group from a low-profile outfit to the poster child for misuse of certain types of technology. Let me recap these:
- SNOWDEN. Edward Snowden’s oath to keep information secret was broken with his notable data dump. Some of these 2013 materials provided sufficient information about specialized software and services to create or release a desire to know more about the market segment.
- CITIZENS LAB. In 2016 Citizen’s Lab kicked off its coverage of the specialized software niche with “The Million Dollar Dissident: NSO Group’s iPhone Zero Days Used against a UAE Human Rights Defender.”
- PITCHING NSO. In 2017, Francisco Partners’ mounted an effort to sell NSO Group for an asking price of around $1 billion. Venture and finance types perked up their ears. Some asked, “What’s this specialized service cyber software?”
- BOOK. In 2019, Shoshana Zuboff published “The Age of Surveillance Capitalism,” which provided a “name” to some of the specialized software functions.
Reporters, activists, researchers, academics, and companies not previously aware of the specialized service sector are now chasing information. Unlike some commercial market sectors, funds are available. The appetite for advanced software and services is growing.
Now back to the company named associated by some with an insect. What impact will the Intercept write up have. I don’t know, but I have three ideas:
First, the company will become a subject of interest for some; for example, an investigative reporter working on a book about the specialized service sector.
Second, non-LE and intel-related organizations will express an interest in licensing the software and gaining access to the firm’s database and other technology. (Voyager Labs has explored selling its software for “marketing.”)
Third, the company’s willingness to market its products and services more aggressively may be reduced. Shadowdragon advertised for a marketing professional, presumably to support the company’s sales efforts. One of the firm’s senior managers posts on LinkedIn in order to express support for certain activities and retain visibility in that Microsoft owned service.
From my narrow point of view, some information should not be exposed to the public; for example, the Snowden dump. And some of the marketing activities of specialized service providers should be wound back to the low profile activities of the pre-911 era.
Unfortunately it may be too late. Commercial success may be more important than creating solutions which support LE and intelligence operations. Today anyone can enjoy useful tools. Check out Hunchly OSINT or Maltego. Explore what these tools can do.
Will Shadowdragon become collateral damage as a consequence of NSO Group?
Stephen E Arnold, September 23, 2021
Apple, Facebook, and an Alleged Digital Trade for a Contentious Product
September 23, 2021
I read “Apple Threatened Facebook Ban over Slavery Posts on Instagram.” I have nothing but respect for the BBC, Brexit, and, of course, the Royals. I also believe everything I read online. (Doesn’t everyone?) Against this background, this BBC slavery write up is interesting indeed.
I read this passage twice to make sure I was getting the message:
Apple threatened to remove Facebook’s products from its App Store, after the BBC found domestic “slaves” for sale on apps, including Instagram, in 2019. The threat was revealed in the Wall Street Journal’s (WSJ) Facebook Files, a series of reports based on its viewing of internal Facebook documents.
Okay. Slave trade. Facebook. Info from “internal Facebook documents.”
Here’s another passage I circled with my trusty red Sharpie Magnum marker:
The trade was carried out using a number of apps including Facebook-owned Instagram. The posts and hashtags used for sales were mainly in Arabic, and shared by users in Saudi Arabia and Kuwait.
Okay. Arabic. Saudi Arabia. Kuwait.
And the Sir Gawain in this matter? China-compliant Apple.
It [Murdoch owned Wall Street Journal] said the social media giant only took “limited action” until “Apple Inc. threatened to remove Facebook’s products from the App Store, unless it cracked down on the practice”.
I hear the digital French Foreign Legion’s tune Le Boudin. Do you?
And the good news? The BBC stated:
In its June 2020 response to these, Facebook wrote: “Following an investigation prompted by an inquiry from the BBC, we conducted a proactive review of our platform. We removed 700 Instagram accounts within 24 hours, and simultaneously blocked several violating hashtags.” The following month the company said it removed more than 130,000 pieces of Arabic-language speech content related to domestic servitude in Arabic on both Instagram and Facebook. It added that it had also developed technology that can proactively find and take action on content related to domestic servitude – enabling it to “remove over 4,000 pieces of violating organic content in Arabic and English from January 2020 to date”.
Interesting indeed. Slavery. Facebook. Social media. Prompt action documented. Apple the pointy end of the stick for justice. Possible vacation ideas for some. The BBC. And more. Quite a write up.
Stephen E Arnold, September 23, 2021
Google and Its Informed Approach to Compensating Employees: Just Pay Less
September 23, 2021
Google does have unethical business practices, including violating user privacy, discriminating treat of ethnic minorities and women in its hiring activities and work environment, and attempting to establish a business relationship with China. Another infraction to add to Google’s growing list is: “Google Underpaid Thousands Of International ‘Shadow Workers,’ Violating Labor Laws Around The World, Report Reveals” says Business Insider.
Google employs over 900 temporary workers in Poland, France, Netherlands, Germany, India, Ireland, and the UK. Pay-parity laws in Asia and Europe require companies to pay full-time and temporary workers the same wage if they essentially do the same job. Google underpaid its workers in violation of these laws.
Google’s compliance department discovered the error, but did not want to bring attention to the issue. Instead Google raised wages for new employees in order to avoid financial, reputational, and legal problems. Google is still attempting to save face:
“While the team hasn’t increased the comparator rate benchmarks for some years, actual pay rates for temporary staff have increased numerous times in that period,’ Spyro Karetsos, Google’s chief compliance officer, said in a statement to Insider. Most temporary staff are paid significantly more than the comparator rates.
‘Nevertheless, it’s clear that this process has not been handled consistently with the high standards to which we hold ourselves as a company,’ he added. ‘We’re doing a thorough review, and we’re committed to identifying and addressing any pay discrepancies that the team has not already addressed. And we’ll be conducting a review of our compliance practices in this area. In short, we’re going to figure out what went wrong here, why it happened, and we’re going to make it right.’”
Google certainly has the funds to make it right. Google should be treating all of its employees, no matter their status, equality. If Google had admitted the mistake, they would have looked like the bigger person, been criticized or praised, then the world would have forgotten the incident. Now Google can add this to their list of personnel management achievements.
High school behavior? Intentional disregard for employee rights? We don’t know.
Whitney Grace, September 23, 2021
Alphabet Spells Out YouTube Recommendations: Are Some Letters Omitted?
September 23, 2021
I have been taking a look at Snorkel (Stanford AI Labs, open source stuff, and the commercial Snorkel.ai variants). I am a dim wit. It seems to me that Google has found a diving partner and embracing some exotic equipment. The purpose of the Snorkel is to implement smart workflows. These apparently will allow better, faster, and cheaper operations; for example, classifying content for the purpose of training smart software. Are their applications of Snorkel-type thinking to content recommendation systems. Absolutely. Note that subject matter experts and knowledge bases are needed at the outset of setting up a Snorkelized system. Then, the “smarts” are componentized. Future interaction is by “engineers”, who may or may not be subject matter experts. The directed acyclic graphs are obviously “directed.” Sounds super efficient.
Now navigate to “On YouTube’s Recommendation System.” This is a lot of words for a Googler to string together: About 2,500.
Here’s the key passage:
These human evaluations then train our system to model their decisions, and we now scale their assessments to all videos across YouTube.
Now what letters are left out? Maybe the ones that spell built-in biases, stochastic drift, and Timnit Gebru? On the other hand, that could be a “Ré” of hope for cost reduction.
Stephen E Arnold, September 23, 2021