Snorkel: Now Humans Are a Benefit?
November 23, 2022
Snorkel emerged from Stanford University’s AI lab. Some at the Google are ga-ga over Snorkel’s approach to reducing the cost of creating training sets for machine learning. If you are not paying attention to the expense of training models the old-fashioned way, when humans do the work, months or years of effort are required. Then — surprise — after operating in the real world for six months (plus or minus depending on the use case), the model has to be retrained.
Snorkel wants to get subject matter experts to build a training set one time. Then the numerical recipes will harvest additional information and automatically update the training set. Imagine better, faster, cheaper. Well, that’s the theory. Thus the entire AI industry push for finding short cuts to deal with the need for building training sets for initial model training and the work needed to make sure the model does not drift off into craziness. (I won’t mention the name of any search vendors, but a number of these outfits have performed oblation for their VC gods. Why? The results of the user’s query returned garbage. Confusing the information in a PowerPoint pitch with returning relevant and precise results for a user’s query is a bit like resolving the conflicts between Newtonian and quantum physics.)
I read “AI Startup Snorkel Preps a New Kind of Expert for Enterprise AI.” My immediate reaction was a question, “Why didn’t Google buy the company?” Hmmm. Now Snorkel is going to push to be a commercial success, perhaps like DeepDyve, an outfit which used or uses Snorkel technology.
The write up says:
Snorkel’s Data-centric Foundation Model Development, as the offering is called, is an enhancement to the startup’s flagship Snorkel Flow program. The new features let companies write functions that automatically create labeled training data by using what are called foundation models, the largest neural nets that exist, such as OpenAI’s GPT-3. The new functions in Snorkel Flow let a person who is a domain expert but not a programmer create a workflow that will then automatically generate labeled data sets that can be used to train the foundation programs for specific tasks.
The base technology emerged from projects guided in part by Christopher Ré. The work goes back more than a decade. Snorkel itself has been a start up for several years.
Smart software is getting a lot of tire kicking action by large companies. My hunch is that Snorkel wants to sell its methods to the firms just now having a bean counter come to a meeting and saying, “Have you taken a look at how much money our AI teams need to retrain our models?”
Then a whiz kid — possibly a graduate of Stanford — says, “Get Snorkel!”
Well, that’s my hunch. Will the models avoid the horrible fate of self immolating smart software which just gets stuff wrong? Probably not. But the PowerPoints and Zoom presentations will explain that Snorkel does not go “under water.” Snorkel lets an apoplectic accountant breathe somewhat more easily until the next quarterly analysis of smart software expenses.
Stephen E Arnold, November 23, 2022
AI: Black Boxes ‘R Us
November 23, 2022
Humans design and make AI. Because humans design and make AI, we should know how they work. For some reason, humans do not know how AI works. Motherboard on Vice explains that, “Scientists Increasingly Can’t Explain How AI Works.” AI researchers are worried that AI developers focus too much on the end results of an algorithm than how and why it arrives at said results.
In other words, developers cannot explain how an AI algorithm works. AI algorithms are built from layers and layers of deep neural networks (DNNs). These networks are designed to replicate human neural pathways. They are almost like real neural pathways, because neurologists are unaware of how the entire brain works and AI developers do not know how AI algorithms work. AI developers are concerned with the inputs and outputs, but the in-between is the mythical black box. Because AI developers do not worry about how they receive the outputs, they cannot explain why they receive biased, polluted results.
“‘If all we have is a ‘black box’, it is impossible to understand causes of failure and improve system safety,’ Roman V. Yampolskiy, a professor of computer science at the University of Louisville, wrote in his paper titled “Unexplainability and Incomprehensibility of Artificial Intelligence.” ‘Additionally, if we grow accustomed to accepting AI’s answers without an explanation, essentially treating it as an Oracle system, we would not be able to tell if it begins providing wrong or manipulative answers.’”
It sounds like the Schrödinger’s cat of black boxes.
Developers’ results are driven by tight deadlines and small budgets so they concentrate on accuracy over explainability. Algorithms are also (supposedly) more accurate than humans, so it is easy to rely on them. Making the algorithms less biased is another black box, especially when the Internet is skewed one way:
“Debiasing the datasets that AI systems are trained on is near impossible in a society whose Internet reflects inherent, continuous human bias. Besides using smaller datasets, in which developers can have more control in deciding what appears in them, experts say a solution is to design with bias in mind, rather than feign impartiality.”
Couldn’t training an algorithm be like teaching a pet to do tricks with positive reinforcement? What would an algorithm consider a treat? But did a guy named Gödel bring up incompleteness? Clicks, clicks, and more clicks.
Whitney Grace, November 23, 2022
A Trifecta for Meta, TikTok, and Twitter in Kenya
November 23, 2022
Once again, social media companies show their disdain for local laws and information integrity. Rest of World reports, “Facebook and Instagram Ran Ads Violating Kenyan Election Law, New Report Reveals.” Furthermore, according to the Mozilla Foundation report, Meta, Twitter, and TikTok failed to moderate harmful posts amid the Kenyan general election in August. Journalist Andrew Deck writes:
“Kenyan law states political candidates cannot campaign in the 48 hours before an election day. Candidates for both major political parties did just that, with paid promotions on Facebook and Instagram, which are both owned by Meta. Meta itself requires advertisers to abide by these blackout periods. Some ads from the opposition Azimio la Umoja party reached as many as 50,000 impressions and one gubernatorial candidate alone ran some 17 violating ads. … The porousness of moderation filters during this time contributed to what [Mozilla researcher Odanga Madung] calls a ‘post-election twilight zone,’ the report said. Despite public commitments to ramp up moderation resources before Kenyans headed to the polls, Meta, Twitter, and TikTok all saw breaches in their moderation systems, according to the report. In the days after the polls closed on August 9, election rumors on social media were exacerbated by the release of 43,000 polling station results publicly by the country’s Independent Electoral and Boundaries Commission (IEBC). Political parties and media companies released their own tallies of these votes, leading to conflicting declarations of the winner. Breaches included the circulation of misleading electoral tallies by opposing political parties and conspiracy theories about election fraud.”
What an interesting matter. See the article for more election chicanery that made it unchallenged onto social media. Meta, TikTok, and Twitter all insist they did their best to uphold regulations and label misinformation. Madung, however, believes they did not adequately test their procedures within Kenya. That seems like a sound conclusion. Just how long will these companies’ negligence contribute to election turmoil in countries around the globe?
Cynthia Murrell, November 23, 2022
OSINT: HandleFinder
November 22, 2022
If you are looking for a way to identify a user “handle” on various social networks, you may want to take a look at HandleFinder. The service appears to be offered without a fee. The developer does provide a “Buy Me a Coffee” link, so you can support the service. The service accepts a user name. We used our favorite ageing teen screen name ibabyrainbow or babyrainbow on some lower profile services. HandleFinder returned 31 results on our first query. (We ran the query a second time, and the system returned 30 results. We found this interesting.)
The services scanned included Patreon, TikTok, and YouTube, among others. The service did not scan the StreamGun video on demand service or NewRecs.
In order to examine the results, one clicks on service name which is underlined. Note that once one clicks the link, the result set is lost. We found that the link should be opened in a separate tab or window to eliminate the need to rerun the query after after each click. That’s how one of my team discovered the count variance.
When there is no result, the link in HandleFinder does not make this evident. Links to ibabyrainbow on Instagram returned “Page not found.” The result for Linktr.ee returned the Linktr.ee page of links, which means more clicking.
If one is interested in chasing down social media handles, you may want to check out this service. It is promising and hopefully will be refined.
Stephen E Arnold, November 22, 2022
Management Follies: High School Science Clubism at Scale!
November 22, 2022
Several years ago, I floated the idea of a “high school science club management method.” A big time “real” journalist said, “That’s wrong. There is a bro mafia management style.”
I said, “Yeah, you are right. I paid the coffee bill and vamoosed. I wanted to bide my time until I had a few rock solid cases of the HSSCMM or high school science club management method. I am speaking from experience because I was in my high school science club. I was one of the three musketeers who liberated the morning public address announcement with Jerry Lee Lewis’ “Great Balls of Fire.”
Yep, a half century ago a band of merry teens (both young men and a couple of young women) met weekly to do science like the PA address thing. Why? I have given this great thought over the last 64 years, and here are the three main reasons we pursued our high value projects.
- We believed that we were special because we earned top grades, accrued honors (one of our club’s members published a short article in a peer-reviewed astronomy journal and yours truly had a story published in the St. Louis Post Dispatch for which I was paid $10. Net net: We thought we were really smart.
- We believed that we knew exactly what to do in any circumstance life in a Midwestern corn and factory town could plop in our path. We have confidence, which I now understand is near fatal hubris.
- We conducted ourselves as if our world was the one, true world: Math, science, logic, etc. etc.
Over the years, I have interacted with a wide range of companies. I have watched the HSSCMM move from the weird world of immature and naïve teens into the mainstream of business, leadership, and — dare I say it — techno worship.
Now back to the real journalist dismissing HSSCMM. That bright individual was incorrect. The HSSCMM is the driving force of several fascinating business dramas now unfolding in real time. These are real people (still locked in the teen year immaturity hormone rush), and having a profound impact on families, children, culture, and the way people perceive the world.
Let’s take a look at the three modern dramas, which if we were in Athens in 440 BCE, would be the stuff for the modern version of Oedipus at Colonus called “Dorks of Silicon Valley” or a bop variant of Women of Trachis named “Bros of Business.”
The first is the Twitter hellscape and the space Karen calling the shots with HSSCMM spins. A Silicon Valley infused article in the Verge provides basic information of a consequence of an informed management decision. To keep the story short, lots of people are just quitting. So the lead Tweeter is shutting down the company for a cou8ple of days. Brilliant cost savings and publicity in one fell swoop. That’s a hallmark of the HSSCMM: Act without considering consequences.
The second is the FTX thing. The estimable Bloomberg has covered the story with slightly more thoroughness than the news organization did its story about fiddled motherboards. I am not going to review this case example because I am not sure about what’s fact and what’s fiction. It does seem as though adding some chemical compounds to a mathematical experiment produced consequential results. Ah, who lost their retirement savings? Too bad. The events appear to follow a path my science club took when we poured calcium carbine and concentrated hydrochloric acid into a lab table drain. Yep, someone ignited the resulting gas. Exciting and no one got expelled.
The third is the Amazon termination of a mere 10,000 humanoids. You can get a sense of what goes into a brilliant and logical decision to cut costs in a recent CNBC report. The reality is the science club or Amazon carpetland. Those not walking on squishy floor coverings cannot be expected to understand. A basic precept of the HSSCMM is that no one in the science club really, really cares.
I am tempted to step back and offer a few thoughts about the broader implications of allowing the HSSCMM to dominate a major part of current business. I won’t. If you read some of the 12,000 posts we have published in Beyond Search since 2008, you can figure out that the fixes are going to be difficult because no one wants to confront the naïve teener mentality which has implemented businesses which leave scorched earth and burned lives.
Yapping and hand waving won’t solve the problem. I am too old to care. I won’t even remind the big time “real” journalist that it is HSSCMM, not bro stuff.
Stephen E Arnold, November 22, 2022
Apple: Curating with Care. What Are the Odds?
November 22, 2022
No one likes ads. They are a necessary evil to keep services free. Unfortunately, users are experiencing more ads than their desired content. Even worse, Apple has transformed its App Store into a giant ad scam hole. Ars Technica shares Apple users’ frustration about the latest iOS update: “Why The App Store’s Tone-Deaf Gambling Ads Make Me Worry About Apple.”
The newest iOS update included important security upgrades, bug fixes, and new features. The App Store’s ad services framework was also included in the upgrade, but instead of repairing problems, the store is now an obnoxious purveyor of irrelevant ads. The store is flooded with ads advertising gambling and get-rich-quick crypto scams. What is even worse is that these ads are playing next to games intended for kids.
Apple is aware of the problem and shut down the worst of the ads, but long-term problems are still an issue.
The bigger question is that Apple is no longer differentiating itself from its rivals. Unlike Microsoft, Facebook, and Google, Apple used to make most of its revenue from hardware sales. Its hardware sales are down, so Apple is compensated for the lower profit margins. It comes at the cost of users’ confidence that their Apple products protected their privacy more than others.
“That’s why Apple’s excursions into the ad business and the increased importance of the Services division to Apple’s continued growth worry me. Not because I think Apple’s products will become unusable or because I think the iPhone or Apple TV home screen is going to become dominated overnight by Roku-style half-page ads, but because I think that the pressure for Apple to degrade the experience for users and developers in the name of expanding its ad business will gradually increase as Apple tries to satisfy shareholders looking for perpetual growth.”
Apple is regarded as a high-quality product, a cut above a basic PC. The expensive price tag and the better privacy OS augments that reputation, but if Apple becomes more like its rivals it will not long per elitist. UGH!
Whitney Grace, November 22, 2022
SocialFi, a New Type of Mashup
November 22, 2022
Well this is quite an innovation. HackerNoon alerts us to a combination of just wonderful stuff in, “The Rise of SocialFi: A Fusion of Social Media, Web3, and Decentralized Finance.” Short for social finance, SocialFi builds on other -Fi trends: DeFi (decentralized finance) and GameFi (play-to-earn crypto currency games). The goal is to monetize social media through “tokenized achievements.” Writer Samiran Mondal elaborates:
“SocialFi is an umbrella that combines various elements to provide a better social experience through crypto, DeFi, metaverse, NFTs, and Web3. At the heart of SocialFi aremonetization and incentivization through social tokens. SocialFi offers many new ways for users, content creators, and app owners to monetarize their engagements. This has perhaps been the most attractive aspect of SocialFi. By introducing the concept of social tokens and in-app utility tokens. Notably, these tokens are not controlled by the platform but by the creator. Token creators have the power to decide how they want their tokens to be utilized, especially by fans and followers.”
This monetization strategy is made possible by more alphabet soup—PFP NFTs, or picture-for-proof non fungible tokens. These profile pictures identify users, provide proof of NFT ownership, and connect users to specific SocialFi communities. Then there are the DAOs, or decentralized autonomous organizations. These communities make decisions through member votes to prevent the type of unilateral control exercised by companies like Facebook, Twitter, and TikTok. This arrangement provides another feature. Or is it a bug? We learn:
“SocialFi also advocates for freedom of speech since users’ messages or content are not throttled by censorship. Previously, social media platforms were heavily plagued by centralized censorship that limited what users could post to the extent of deleting accounts that content creators and users had poured their hearts and souls into. But with SocialFi, users have the freedom to post content without the constant fear of overreaching moderation or targeted censorship.”
Sounds great, until one realizes what Mondal calls overreach and censorship would include efforts to quell the spread of misinformation and harmful content already bedeviling society. To those behind SocialFi have any plans to address that dilemma? Sure.
Cynthia Murrell, November 22, 2022
Is This FTX Math?
November 21, 2022
I recall hearing or reading that the top dog of Aurora, the FTX affiliated entity, thought that running a crypto outfit required basic math skills. I read “FTX Owes more than $3 Billion to Top 50 Crypto Creditors.” The write up points out:
Roughly 1 million customers unable to withdraw funds from crypto currency exchange
Okay, let’s do some modern day basic math. Follow along:
- 50 kind hearted and understanding creditors
- $3 billion in US dollars or a suitable equivalent like Teslas or designer bags
- One Enron experienced super manager.
Now the math:
50 times $3 billion equals 20 years
I will get a middle school student to check my math. I may be off a few years with the prison sum.
Stephen E Arnold, November 21, 2022
Does Medium Promote Stolen Software?
November 21, 2022
I read “Bigasoft Total Video Converter 6.4.2.8118 Crack with Serial Key 2023.” The main idea of the write up is that a reader of this Medium article can steal intellectual property. The tip off for a human subject matter expert reviewing content for appropriateness or smart software jazzed on Snorkelesque magic would probably note the word “crack” and the phrase “with Serial Key 2023.” The fact that this article appears on Medium is probably something will annoy the developers of Bigasoft’s video converter. My hunch is that a legal eagle may want to call this Medium oversight to the attention of a responsible adult at the zippy alternative to an old school magazine. What does this somewhat obvious invitation to steal software say? I quote:
Bigasoft Total Video Converter Keygen is compatible with all versions of Windows and runs smoothly on Mac….After reading the installation instructions below, you may install this application since it is effortless. You get TipuCrack from it….
Yeah, this seems clear. Am I missing something that makes this type of theft advocacy okay? I hope my link goes dead because that shows a modicum of ethical behavior.
Stephen E Arnold, November 21, 2022
Estonia and e-Residency
November 21, 2022
I have been to Estonia a couple of time. Once I visited in the summer. Another time I visited in February. Here’s a tip: “Leaves of Grass” weather is preferable in my opinion.
I mention Estonia because I noted a link to the Estonian government’s e-Residency information. You can find the basics at “Become and E-Resident.”
The main idea is that one can join Estonia’s digital nation. E-Residency is open to people from other countries. The idea is that the business would be “location independent” and the company would be an EU outfit.
The benefits include:
- Grow your business remotely
- Minimized bureaucracy (keep in mind that this is an EU company within a Baltic state with a Russian border)
- Joining an international community.
There are nominal fees, probably less than US$200, and a background checking process.
The idea is an interesting one. However, the e-Residency does not appear to include one of those “golden passports” available from some countries.
Are there downsides? A few, for example:
- Explaining to a US tax authority what’s going on
- Anticipating how the program will evolve; for example, laws passed in Estonia going forward
- Dealing with litigation in the US, EU, and elsewhere
- Resolving issues arising from payment to vendors and collecting money from customers.
If this approach to business appears attractive, check out the Estonia government’s Web site.
Stephen E Arnold, November 21, 2022
-
- Subscribe to Beyond Search
Feature archive
News archive
-
