Monday, September 30, 2019


There’s a large bias downside: If they use US data only their understanding of North Korea or China or Iran will be skewed.
The Intelligence Community needs to be willing to ingest data from any source to unleash the power of artificial intelligence (AI), and to provide decision-makers with rapid, in-depth insights they can use, says Sue Gordon.
“We need to be able to use more data, and trust the data we use. That is a really hard thing for us,” said Gordon
… The value of AI, she said, lies in its ability to rapidly sift through mountains of data — “to use the bigness of the data” — to find key information, patterns of activities and behavior, and to “understand trends in ways that we can’t see.”
… Specifically, she said the IC has to change its internal “data ownership” culture. “We’ve gotta get away from people thinking that because they were responsible for collecting it, they are the ones who get to determine how it’s used.” In addition, she said, a change is needed so that “the people who are collecting data don’t get to decide how it is collected; it has to be collected with use in mind.”




Food for thought.
How to Keep Company IoT Networks Secure
… Over 20 percent of companies reported a data breach or cyberattack in 2018 that could be traced back to unsecured IoT devices, according to a Ponemon Institute Survey.
This article will focus on the approaches that help build a better Internet of Things security system and prevent a destructive breach.




Gee, nobody else complies – why should I?
Why the fate of online accessibility may rest with a Domino's Pizza lawsuit
When the Americans with Disabilities Act was passed in 1990, the internet wasn't yet a critical part of our lives. We weren't connected to it 24/7, and we didn't rely on it to handle everyday tasks and provide a steady stream of information.
… Guillermo Robles sued the restaurant in 2016 after he wasn't able to order a custom pizza from its website and app, even while using screen-reading technology. The appeals court in the Ninth Circuit sided with Robles, ruling the ADA does in fact apply to websites and apps of businesses with physical locations. Now, the company is asking the Supreme Court to review the case. It'll decide whether to do so in the fall.
… One reason many businesses lag behind when it comes to accessibility is a lack of enforcement, Coelho said. In 2010, the Justice Department said it was going to put out web accessibility regulations, but it ultimately ditched that plan in 2017. Courts have ruled the ADA applies online, but that hasn't been enough to convince businesses across the board to implement accessibility initiatives.




First an oil processing facility in Saudi, next something much closer to me?
The Drone Databook – study of military drone capabilities
The Center for the Study of the Drone at Bard College – “The Drone Databook [September 2019] is an in-depth survey of the military drone capabilities of over 90 countries around the globe that are known to possess and operate unmanned aircraft. This study includes information about each country’s active drone inventories, drone units, training programs, active acquisition and development efforts, infrastructure, drone exports, and operational experiences.”
“…Once a novelty, drones have become standard military equipment, spawning a global network of units, bases, and test sites. Battlefields in Ukraine, Syria, and Yemen, as well as zones of geopolitical conflict such as the Persian Gulf and the East China Sea, are increasingly crowded with drones of varying size and sophistication. Whether they are used for intelligence gathering, aerial strikes, artillery spotting, or electronic warfare, drones are a leading contributor to the changing character of modern war. The Drone Databook is a study of military drone capabilities. It is comprised of profiles of 101 countries in seven regions – Asia and Oceania, Eurasia, Europe, Latin America, the Middle East and North Africa, North America, and Sub-Saharan Africa – as well as two appendixes that address military drone infrastructure around the globe and the technical specifications of more than 170 drones currently in use by these countries. The Databook evaluates the military drone capabilities of each country in terms of six categories: inventory and active acquisition programs, personnel and training programs, infrastructure, operational experience, aircraft research and development programs, and exports…”




Will AI train lawyers to be better listeners and story tellers?
Ok, Google, Will Artificial Intelligence Replace Human Lawyering?
Oseid, Julie A. and Vorenberg, Amy and Koenig, Melissa Love, Ok, Google, Will Artificial Intelligence Replace Human Lawyering? (2019). 102 Marquette Law Review 1269 (2019); U of St. Thomas (Minnesota) Legal Studies Research Paper No. 19-13; Marquette Law School Legal Studies Paper No. 19-13. Available at SSRN: https://ssrn.com/abstract=3449500 – “Will Artificial Intelligence (AI) replace human lawyering? The answer is no. Despite worries that AI is getting so sophisticated that it could take over the profession, there is little cause for concern. Indeed, the surge of AI in the legal field has crystalized the real essence of effective lawyering. The lawyer’s craft goes beyond what AI can do because we listen with empathy to clients’ stories, strategize to find that story that might not be obvious, thoughtfully use our imagination and judgment to decide which story will appeal to an audience, and creatively tell those winning stories. This article reviews the current state of AI in legal practice and contrasts that with the essence of exclusively human lawyering skills—empathy, imagination, and creativity. As examples, we use three Supreme Court cases to illustrate these skills.”




Are these communications secure?
LAW FIRMS AND THEIR CLIENTS ARE INCREASINGLY INTERACTING VIA TEXT OR CHAT MESSAGES
As email goes the way of phone calls—increasingly perceived as a formal communication channel requiring considerable time and effort—many attorneys are finding themselves communicating with clients through text messages or chat platforms.
… "To the extent we receive material information in a non-email communication from a client, the practice of our firm is to retain that message," he explains.
But his firm does not actively manage the way he or its other attorneys interact with clients, and there are signs that it’s not alone. The 2019 LTN Law Firm Tech Survey found that only one-third of its 36 law firm respondents had an official policy for collecting non-email sources of electronic data, underscoring how most don’t supervise their attorneys’ communications.



Sunday, September 29, 2019


What are the minimum requirements for a disclosure order?
The U.S. and UK governments are expected to sign a treaty in October that will force social media platforms based in either of the countries to “disclose encrypted messages from suspected terrorists, paedophiles and other serious criminals” to police in the other, according to the Times of London.
… According to the paper, Patel said UK authorities’ hands are currently tied by arrangements that keep transnational data-sharing to emergencies and a slow-moving treaty process:
At present the security services are only able to obtain data if there is a need for an “emergency disclosure” due to an imminent threat to life. The police and prosecutors can also request data under the “mutual legal assistance” treaty but the process is highly bureaucratic and can take up to two years.
Under the new treaty, the police, prosecutors and the security services can submit requests for information to a judge, magistrate or “other independent authority”. The process will be overseen by the investigatory powers commissioner.
Under the terms of the proposed arrangement, both governments will agree not to investigate each others’ citizens. The U.S. won’t be able to use data obtained from companies based in the UK in death penalty cases unless UK authorities have explicitly given permission to do so. Bloomberg confirmed news of the data sharing agreement as well.
It’s not clear whether the proposed arrangement actually requires companies to build backdoors into their encrypted products, something that law enforcement and intelligence agencies have been demanding for years, but which has been resisted by tech firms.




A bit snarky, don’t you think?
Mike Masnick writes:
California is inching ever closer to having its very problematic privacy law take effect. As we’ve noted, while good privacy legislation would be desirable, this is not it. Indeed, this law is woefully undercooked by design. If you don’t remember, the process by which we got here dictated terrible results. A wealthy real estate developer, Alastair Mactaggart, decided that he was going to “fix” internet privacy, by putting a truly bad proposal regarding internet privacy to a public vote, using California’s somewhat horrific public referendum system — that allows for the public to effectively modify California’s constitution by popular vote.
Read more on TechDirt.
[From the article:
But, here's the thing, after agreeing to pull that referendum from the ballot, Mactaggart has now announced that he's bringing it back for the next ballot.




Legal creep?
Odia Kagan of Fox Rothschild writes:
The Danish Data Protection Authority has changed its position regarding the legal basis for posting pictures online under the General Data Protection Regulation (GDPR). Rather than a distinction between “situational” and “portrait” pictures, Datatilsynet now requires a case-by-case analysis.
Read more on Fox Rothschild.




I’ve come to expect poor reporting. Did anyone notice that this was implemented without notice? That it does not scan the Internet as the school board seems to claim. And did they check that claim of preventing suicides?
Safety versus privacy: Williamson County School introduces new 'Gaggle' student surveillance program
After the Williamson County School District implemented a threat surveillance computer program, some parents are concerned about protecting students’ privacy.
… WCS Superintendent Jason Golden announced the use of the online safety program at a school board work session in August. Without giving details due to federal family protection laws, Golden explained the new program had already “paid for itself” this year by identifying a threat of student self-harm, allowing district officials to intervene.
According to the Gaggle distributor, during the 2018-19 school year, the program helped school districts save 722 students from carrying out an act of suicide.
… Gaggle operates using a mathematical algorithm to identify high risk words and phrases when students are logged into the district’s server.
Gaggle scans WCS student accounts in G Suite for Education within the district for inappropriate or concerning words and images.
Systems monitored by Gaggle in WCS include Gmail, Google Docs, Google Drive and other programs using myplace.wcs.edu accounts provided to students.
The district says Gaggle is aimed at protecting students on the internet, and it's the district role to do so, according to federal law.
… Gaggle does not monitor student internet use.




Just because I found it…
The Ethics of Artificial Intelligence
Undergraduate Thesis




I’ll definitely use Dilbert’s explanation of BlockChain.



Saturday, September 28, 2019


Who do you want to win and by how much?
Researchers easily breached voting machines for the 2020 election
The voting machines that the US will use in the 2020 election are still vulnerable to hacks. A group of ethical hackers tested a bunch of those voting machines and election systems (most of which they bought on eBay). They were able to crack into every machine, The Washington Post reports. Their tests took place this summer at a Def Con cybersecurity conference, but the group visited Washington to share their findings yesterday.
A number of flaws allowed the hackers to access the machines, including weak default passwords and shoddy encryption. The group says the machines could be hacked by anyone with access to them, and if poll workers make mistakes or take shortcuts, the machines could be infiltrated by remote hackers.




Is this an overreaction?
New federal rules limit police searches of family tree DNA databases
The U.S. Department of Justice (DOJ) released new rules yesterday governing when police can use genetic genealogy to track down suspects in serious crimes—the first-ever policy covering how these databases, popular among amateur genealogists, should be used in law enforcement attempts to balance public safety and privacy concerns.
… But these searches also raise privacy concerns. Relatives of those in the database can fall under suspicion even if they have never uploaded their own DNA. (One study found that 60% of white Americans can now be tracked down using such searches .) And even those who have shared their DNA may not have given informed consent to allow their data to be used for law enforcement searches.
… The policy says “forensic genetic genealogy” should generally be used only for violent crimes such as murder and rape, as well as to identify human remains. (The policy permits broader use if the ancestry database’s policy allows such searches.) Police should first exhaust traditional crime solving methods, including searching their own criminal DNA databases.
… The policy also bars police from using a suspect’s DNA profile to look for genes related to disease risks or psychological traits. [??? Bob]




Lawyers for the defense?
Five Key Considerations to Developing Defensible AI
In a recent survey published by Forbes, 91% of enterprises said they expect AI to deliver new business growth in the next five years.
The rapidly expanding implementation of artificial intelligence in society means that, inevitably, lawyers will be faced with putting artificial intelligence on the witness stand.
… What should companies implementing AI solutions do now to ensure that they can effectively defend their use of AI when the machine becomes the witness? This article offers some thoughts on developing AI in a defensible way in order to be prepared should the AI system itself become the focal point of a lawsuit.
Pay Attention to the Data Used to Train Your AI
Understand the Black Box
Develop AI in an Ethically Responsible Way
Be Mindful of Privacy Regulations
Always Look to Improve Results




Worth a look!
Get Your Copy of the Free Practical Ed Tech Handbook
Last Sunday I published the updated 2019-20 version of my popular Practical Ed Tech Handbook.
Learning to Program
Augmented and Virtual Reality
Video creation and flipped lessons



Friday, September 27, 2019


Remember, some hackers can read.
GAO Identifies Significant Cybersecurity Risks in US Electric Grid
The new report released by the Government Accountability Office (GAO) reveals that the nation’s electric grid is becoming more vulnerable to cyberattacks.


(Related)
U.S. Navy to Appoint Cyber Chief Following a Blistering Audit
… The new position is part of a broader effort to improve cybersecurity in the Navy and among its private-sector industry partners, coming after a scathing internal audit earlier this year found that repeated compromises of national-security secrets threatened the U.S.’s standing as the world’s top military power.




For my security students: know the enemy!




Before this one becomes law, lets design a tougher one!
CCPA 2.0? A New California Ballot Initiative is Introduced
On September 13, 2019, the California State Legislature passed the final CCPA amendments of 2019. Governor Newsome is expected to sign the recently passed CCPA amendments into law in advance of his October 13, 2019 deadline. Yesterday, proponents of the original CCPA ballot initiative released the text of a new initiative (The California Privacy Rights and Enforcement Act of 2020) that will be voted on in the 2020 election; if passed, the initiative would substantially expand CCPA’s protections for consumers and obligations on businesses. While the new proposal preserves key aspects the current CCPA statute, there are some notable additions and amendments.




No surprise.
Most Companies Worldwide Are Not Prepared for New Privacy Regulations
As new privacy regulations continue to appear globally, there is mounting evidence to suggest that most organizations – regardless of size or type of business – are unprepared to deal with them in an effective manner. That’s one of the big takeaways from a recent September 2019 report from the Internet Society’s Online Trust Alliance (OTA), which analyzed more than 1,200 privacy statements from organizations around the world to see how well they adhered to new privacy regulations such as the European General Data Protection Regulation (GDPR), the California Consumer Privacy Act (CCPA), and Canada’s Personal Information Protection and Electronic Document Act (PIPEDA).
… In many ways, the enactment of the European GDPR in May 2018 set into motion an entirely new approach to data privacy and data security. This paradigm shift included new thinking about how to manage the enormous flows of data passing into and out of organizations on a daily basis, and how to report this information to customers and users. And it also established the fact that organizations would have to start dealing with enormous fines and penalties for any lapses in data privacy or any data security breaches.




From devices to algorithms.
FDA clarifies how it will regulate digital health and artificial intelligence
… In releasing the guidelines on mobile health software and CDS tools, the FDA is attempting to communicate how it will implement provisions within the 21st Century Cures Act, a law passed in late 2016 that sought to exempt several categories of health software from FDA review. The legislation gave FDA discretion to determine which specific products will fall under its purview.




A not very scientific assumption.
Don’t Fear the Terminator
Artificial intelligence never needed to evolve, so it didn’t develop the survival instinct that leads to the impulse to dominate others
… Takeover by AI has long been the stuff of science fiction. In 2001: A Space Odyssey, HAL, the sentient computer controlling the operation of an interplanetary spaceship, turns on the crew in an act of self-preservation. In The Terminator, an Internet-like computer defense system called Skynet achieves self-awareness and initiates a nuclear war, obliterating much of humanity. This trope has, by now, been almost elevated to a natural law of science fiction: a sufficiently intelligent computer system will do whatever it must to survive, which will likely include achieving dominion over the human race.
To a neuroscientist, this line of reasoning is puzzling. There are plenty of risks of AI to worry about, including economic disruption, failures in life-critical applications and weaponization by bad actors. But the one that seems to worry people most is power-hungry robots deciding, of their own volition, to take over the world. Why would a sentient AI want to take over the world? It wouldn’t.




Perspective. No need to plug in. No need to change batteries.
Photovoltaic-powered sensors for the 'Internet of Things'
By 2025, experts estimate the number of Internet of Things devices—including sensors that gather real-time data about infrastructure and the environment—could rise to 75 billion worldwide.
… MIT researchers have designed photovoltaic-powered sensors that could potentially transmit data for years before they need to be replaced. To do so, they mounted thin-film perovskite cells—known for their potential low cost, flexibility, and relative ease of fabrication—as energy-harvesters on inexpensive radio-frequency identification (RFID) tags.




Perspective.
Internet sector contributes $2.1 trillion to U.S. economy: industry group
The rapidly growing internet sector accounted for $2.1 trillion of the U.S. economy in 2018, or about 10% of the nation’s gross domestic product (GDP), an industry group said on Thursday.
… The study says the internet sector represents the fourth largest sector of the U.S. economy, behind real estate, government and manufacturing.




Something to watch for…
This Website Will Turn Wikipedia Articles Into “Real” Academic Papers
BuzzFeedNews – “The digital product agency MSCHF released a site called M-Journal on Tuesday that will turn any Wikipedia article into a “real” academic article. You can screenshot it, you can cite it — and you can send a link to your teacher. What MSCHF did was republish the entirety of Wikipedia under its own academic journal. If you go over to the site, you can search any Wikipedia article or paste in a link, and it’ll generate a citation that refers to MSCHF’s M-Journal, not Wikipedia…”



Thursday, September 26, 2019


Trying to explain CCPA to my students.
Ready, Set, Sustain: Six Steps Toward CCPA Compliance
The California Consumer Privacy Act (CCPA) is the first major piece of United States privacy legislation, but it won’t be the last. There are already similar bills in the works in Washington, Hawaii, Massachusetts, New Mexico, Rhode Island and Maryland. Introduced on June 28, 2018, the CCPA adopts much of its framework from the European Union General Data Protection Regulation (GDPR) – although there are some subtle differences. For example, the CCPA extends its protections to households and devices, not just individuals, and includes the right to opt-out of the sale of personal information.
… Our research suggests a lot of companies were blindsided by how much time and money it takes to sustain compliance. With less than four months until the California Consumer Privacy Act goes into effect on January 1, 2020, this article provides actionable steps for how to work toward a sustainable compliance program.


(Related)
ALI Data Privacy: Overview and Black Letter Text — Available for Download
Professor Paul Schwartz and I have posted the black letter text of the American Law Institute (ALI), Principles of the Law, Data Privacy. Professor Paul Schwartz and I were co-reporters on the project. Earlier this year, I wrote a post about our completion of the project. According to the ALI press release: “The Principles seek to provide a set of best practices for entities that collect and control data concerning individuals and guidance for a variety of parties at the federal, state, and local levels, including legislators, attorneys general, and administrative agency officials.”




Probably as much insight as we can find.
What's New In Gartner's Hype Cycle For AI, 2019
… Gartner’s definition of Hype Cycles includes five phases of a technology’s lifecycle and is explained here. Gartner’s latest Hype Cycle for AI reflects the growing popularity of AutoML, intelligent applications, AI platform as a service or AI cloud services as enterprises ramp up their adoption of AI.


(Related) Voice is being added to every ‘Internet of Things’ thing.
Alexa's 'Certified for Humans' wants to eliminate smart-home headaches
Amazon wants you to be able to set up your smart home even if you don't know anything about tech.
… To achieve the certification, devices need to use frustration-free setup. This allows Alexa to share your Wi-Fi credentials so you don't have to renter the info. The device needs to allow over-the-air updates to happen in the background. Finally, you need to be able to setup and control the devices with the Alexa app.


(Related)
Ring announces new cameras and a conversational doorbell
At its fall event today, Amazon announced two new versions of its Ring Stick Up security cameras and showed off a way for Alexa to have conversations with people who come to your door via a Ring doorbell.


(Related)
Amazon’s Echo Frames are eyeglasses with Alexa




Perspective.
The Internet Leads in Advertising by a Crazy Wide Margin
… Zenith Media's Advertising Expenditure Forecasts June 2019 covers the world, looks at data from 2007 to 2018, and forecasts it clearly: in a couple of years, digital/online ad spending will account for 52 percent of all the dollars spent to get your attention.
… TV is a distant second place at 27 percent of advertising spending projected for 2021. After that comes an even more precipitous drop. The third place goes to outdoor ads! That means billboards will get more money spent on them than newspapers (6 percent), radio (5 percent), or magazines (3 percent).




PowerPoint is unlikely to rise above ‘tolerable.”
Microsoft launches its AI presentation coach for PowerPoint
A few months ago, Microsoft announced that PowerPoint would soon get an AI-powered presentation coach that could help you prepare for that important next presentation by giving you immediate feedback. Today, the company is launching this new tool, starting with the web version of PowerPoint.
… The new PowerPoint Presentation Coach aims to take the hassle out of practicing. In its current version, the tool looks at three things: pace, slide reading and word choice. Pace is pretty self-explanatory and looks at how fast or slow somebody is speaking. The “slide reading” feature detects when you are simply reading the words from your slides word for word. Nobody wants to sit through that kind of presentation. The “word choice” tool doesn’t just detect how often you say “um,” “ah,” “actually” or “basically,” it also gives you feedback when you are using culturally insensitive phrases like “you guys” or “best man for the job.”



Wednesday, September 25, 2019


Bringing us doom, gloom, and a President who polarizes the country?
Russian Secret Weapon Against U.S. 2020 Election Revealed In New Cyberwarfare Report
The FBI has warned that “the threat” to U.S. election security “from nation-state actors remains a persistent concern,” that it is “working aggressively” to uncover and stop, and the U.S. Director of National Intelligence has appointed an election threats executive, explaining that election security is now “a top priority for the intelligence community—which must bring the strongest level of support to this critical issue.”
With this in mind, a new report from cybersecurity powerhouse Check Point makes for sobering reading. “It is unequivocally clear to us,” the firm warns, “that the Russians invested a significant amount of money and effort in the first half of this year to build large-scale espionage capabilities. Given the timing, the unique operational security design, and sheer volume of resource investment seen, Check Point believes we may see such an attack carried out near the 2020 U.S. Elections.”
… And the most chilling finding is that Russia has built its ecosystem to ensure resilience, with cost no object. It has formed a fire-walled structure designed to attack in waves. Check Point believes this has been a decade or more in the making and now makes concerted Russian attacks on the U.S. “almost impossible” to defend against.
… It’s known and accepted within the U.S. security community that the elections will almost certainly come under some level of attack. But the findings actually point to something much more sinister. A cyber warfare platform that does carry implications for the election—but also for power grids, transportation networks, financial services.
[An alternate link to the report: https://www.intezer.com/blog-russian-apt-ecosystem/


(Related) How can this possibly help?
Facebook promises not to stop politicians’ lies & hate
Facebook confirms it won’t fact check politicians’ speech or block their content if it’s newsworthy even if it violates the site’s hate-speech rules or other policies. This cementing of its policy comes from Facebook’s head of global policy and communication Nick Clegg, who gave a speech today about Facebook’s plans to prevent interference in the 2020 presidential election.




We need security training that is as frequent and as habit forming as dealing with large volumes of email.
Webroot Report: Nearly Half of Employees Confess to Clicking Links in Potential Phishing Emails at Work
BROOMFIELD, Colo., Webroot, a Carbonite (CARB) company, released a report, Hook, Line and Sinker: Why Phishing Attacks Work, that sheds light on psychological factors impacting an individual's decision to click on a phishing email.
While a majority (79%) of people reported being able to distinguish a phishing message from a genuine one, nearly half (49%) also admit to having clicked on a link from an unknown sender while at work. Further, nearly half (48%) of respondents said their personal or financial data had been compromised by a phishing message. However, of that group more than a third (35%) didn't take the basic step of changing their passwords following a breach. Not only is this false confidence potentially harmful to an employee's personal and financial data, but it also creates risks for companies and their data.




Compliance is not always what was intended. (And some companies are smarter than some countries?)
Google refuses to pay publishers in France
Google will not pay press publishers in France to display their content and will instead change the way articles appear in search results, a senior executive said on Wednesday.
The announcement pours cold water on publishers' hopes of obtaining more money from the tech giant for displaying their content under the European Union's new copyright regime, which France was the first to transpose into national law.
… To apply the new copyright rules in France, Google will instead change the way news results appear on its search engine by removing so-called snippets, or short excerpts from the article.
"When the French law comes into force, we will not show preview content in France for a European news publication unless the publisher has taken steps to tell us that's what they want," the tech giant said in a separate blog post.
Hyperlinks and "very short extracts" of press articles are not covered by the neighboring right, meaning Google can display them on the platforms without signing a licensing agreement.
In Germany, where a neighboring right existed prior to the EU directive, many German publishers decided to give Google their content for free after their traffic plummeted when snippets no longer appeared on search results.




My AI can file more patent applications than your AI!
AI at the USPTO
The USPTO has extended its public comment period on the subject of patenting artificial intelligence inventions. Due Date: October 11, 2019 now November 8, 2019.




I suspect they are far too optimistic, but them I’m a pessimist by training.
Explainable AI: Bringing trust to business AI adoption
… At the center of this problem is a technical question shrouded by myth. There's a widely held belief out there today that AI technology has become so complex that it's impossible for the systems to explain why they make the decisions that they do. And even if they could, the explanations would be too complicated for our human brains to understand.
The reality is that many of the most common algorithms used today in machine learning and AI systems can have what is known as “explainability” built in. We're just not using it — or are not getting access to it. For other algorithms, explainability and traceability functions are still being developed, but aren't far out.
Here you will find what explainable AI means, why it matters for business use, and what forces are moving its adoption forward — and which are holding it back.


(Related)
Fiddler raises $10.2 million for AI that explains its reasoning
Explainable AI, which refers to techniques that attempt to bring transparency to traditionally opaque AI models and their predictions, is a burgeoning subfield of machine learning research. It’s no wonder — models sometimes learn undesirable tricks to accomplish goals on training data, or they develop biases  with the potential to cause harm if left unaddressed.
That’s why Krishna Gade and Amt Paka founded Fiddler, a Mountain View, California-based startup developing an “explainable” engine that’s designed to analyze, validate, and manage AI solutions.




I don’t get it. Perhaps I’ll learn eventually?
New AI Systems Are Here to Personalize Learning
… Ahura AI is developing a product to capture biometric data [sic] from adult learners who are using computers to complete online education programs. The goal is to feed this data to an AI system that can modify and adapt their program to optimize for the most effective teaching method.
… Currently, Ahura’s system uses the video camera and microphone that come standard on the laptops, tablets, and mobile devices most students are using for their learning programs.
With the computer’s camera Ahura can capture facial movements and micro expressions, measure eye movements, and track fidget score (a measure of how much a student moves while learning). The microphone tracks voice sentiment, and the AI leverages natural language processing to review the learner’s word usage.
From this collection of data Ahura can, according to Talebi, identify the optimal way to deliver content to each individual.




You want to write right, right?
Grammarly uses AI to detect the tone and tenor of your writing
Nailing the right tone and tenor is of critical importance where writing’s concerned, whether the audiences of said writing are friends, coworkers, or hiring managers. The trouble is, without an extra pair of eyes, it’s rarely easy to know whether your work will have the intended effect.
Fortunately, Grammarly, a developer of cloud-hosted online grammar checking and plagiarism detection tools, has developed a tone detector the company claims can identify subtle contextual clues conveying a range of tempers. Essentially, it taps a battery of hard-coded rules and machine learning algorithms to spot signals in a piece contributing to its tone, including word choice, phrasing, punctuation, and even capitalization.