Will The Google PR Carpet Bombing about AI Deliver Victory?
December 13, 2021
The short answer is, “Yes.” The mass market and niche content marketing is the obvious part of the program. There are some less obvious characteristics which warrant some attention. Run a query for Snorkel. What does one get? Here’s my result page on December 9, 2021:
The pictures are for scuba snorkels. But the first hit is not an advertisement, at least one that an entity compensated the Google to publish. The number one with a bullet is the Snorkel AI company. There you go. That’s either great performance from the circa 1998 algorithm, spectacular SEO, or something of interest to some entity at the Google.
What happens if I run a query for “AI”? Here’s what I saw on December 9, 2021
Amazon bought an ad and linked to its free AI solutions. The number one hit is:
The Google.
So what? Nudging, great SEO, some engineer’s manicured hand on the steering wheel?
I do know that most people have zero idea about smart software. What’s my source for this glittering generality? Navigate to “Survey Suggests 84% of Americans Are Illiterate about AI — So Here’s a Quiz to Test Your Own AI IQ.”
Those nudges, the PR, and the search results may amount to something; for example, framing, reformation, and dissemination of what the Google magical “algorithms” find most relevant. Google wants to win the battle for its approach to really good training data for machine learning. Really wants to win.
Stephen E Arnold, December 13, 2021
What Google Knows about the Honest You
December 10, 2021
I read this quote in a Kleenex story about Google’s lists of popular searches:
“You’re never as honest as you are with your search engine. You get a sense of what people genuinely care about and genuinely want to know — and not just how they’re presenting themselves to the rest of the world.”
The alleged Googler crafting this statement is a data editor. You can read more about the highly selective and unverified Google search trends in “What Google’s Trending Searches Say about America in 2021.”
For me, the statement allows several observations:
- A person acting in an unguarded way reveals information not usually disseminated in “guarded” settings; for example, a job interview
- The word “honest” implies an unvarnished look at the psycho-social factors within a single person
- A collection of data points about the psycho-social aspects of a single person makes it possible to tag, classify, and relate that individual to others. Numerical procedures allow a person or system with access to those data to predict certain behaviors, predispositions, or actions.
Thus, the collection of searches, clicks, and items created by an individual using Google services such as Gmail and YouTube create a palette of color from which a data maestro can paint a picture.
Predestination has never been easier, more automatable, or cheaper to convert into an actionable knowledgebase for smart software. Yep, just simple queries. Useful indeed.
Stephen E Arnold, December 10, 2021
Google and Its Big AI PR Campaign
December 9, 2021
I spotted “DeepMind Says Its New Language Model Can Beat Others 25 Times Its Size.” In my opinion, this is part of the Google play to sail forward with its alleged better, faster, cheaper method of training machine learning models. Most people won’t care or know what’s underway. That’s okay because “information” is now channeled through specific conduits. As long as an answer is good enough or the payoff is big enough for the gatekeepers, the engineering is doing its job.
The write up is happily unaware of this push to use 60 percent or “good enough” accuracy to create the foundation for downstream training set generation. But, oh, boy, is that relaxed supervision great for matching ads. Good enough burns down inventory and it allows machine learning models to be trained on content domains quickly and without the friction imposed by mother hen subject matter experts, rigorous analysis and tuning, and retraining using human intermediated data sets.
Plus, skew, drift, and biases are smoothed out or made to go away. Well, that’s the theory.
The jazzy name Retro is not old school. It is new school. The lessons users will learn will take a long time to understand and appreciate its nuances.
This is a big business play and its accompanying PR campaign is working. Just ask Dr. Timnit Gebru, that is the former Google employee who raised the specter of bias, wonky outputs, and the potential for nudging users down the Googley path.
For another example of Google’s AI PR push, navigate to “DeepMind’s New 280 Billion-Parameter Language Model Kicks GPT-3’s Butt in Accuracy.” Wow, just like quantum supremacy.
Stephen E Arnold, December 9, 2021
Google into Healing
December 9, 2021
How does one explain healing? For the Google the answer is with a Year in Search video. “Google Is All about Healing in its Year In Search Video for 2021” reports:
According to Google Trends, searches on climate change have hit a record high number in 2021. Google’s Year In Search video has captured that with the search of ‘how to help our planet.’ Other positive search examples like ‘ways to help your community’ and ‘how to be yourself’ have also captured the attention of Google and are included in the video with appropriate examples.
Okay, climate change. And Covid. Google even shares some of its user data, according to the write up:
The Year In Search video is accompanied by an interactive site where you can see all of the top global Google searches month-by-month in nine different categories. Of course, you can also check out Google Trends where you can see data for a specific country or region.
What about repairing that relationship with Dr. Timnit Gebru and others who raised questions about Google’s AI methods and motives?
Oh, that’s an unfair question. No healing there. Plus only the really important information is worthy of the Year in Search video.
Stephen E Arnold, December 9, 2021
Smart Software and Cartels: Another View of the Question To Google or Not to Google?
December 7, 2021
I read “A Cartel of Influential Datasets Is Dominating Machine Learning Research New Study Suggests.” The “team” beavering away is an impressive one to the AI big wigs: The University of California at Los Angeles and Google. The findings are interesting. Developers of smart software have relied on widely available datasets. And those datasets can and may have posed a problem. The datasets are not ones that the average computer user will know about or understand what’s in them. But they are available and less expensive than building a collection of data, making sure it is sort of unbiased, benchmarking the dataset, and then deploying it in such a manner than errors or statistical eddies, currents, and drifts are noted and addressed. Wow, that’s a lot of work, and it is expensive. It is just more efficient to use what’s available and trust the “law of big numbers” or the magic of statistical procedures to fill in the potholes.
The problem is that the expensive alternative is a non starter in today’s go go, let’s make money now world. This means that my interpretation of this allegedly objective, peer reviewed, credential bedecked study is different.
Here’s what I think is afoot. The research discredits what most of the companies building machine learning centric solutions is doing. The fix, in my opinion, is Google’s embrace of the principles and practices of the Stanford Artificial Intelligence Laboratory or SAIL. The idea is manifested in Dr. Christopher Ré’s research, the DeepDyve system, and the Snorkel open source software and commercial variants.
The solution is to skip as much of the human involvement in training as possible. Let the downstream system work out the details and fix the pavement in the information superhighway. The Snorkel approach is going to be better in every possible way than using the widely available datasets and a whole lot cheaper than creating training data by hand and then paying quite a few subject matter experts to tune the system.
Net net: My hunch is that Google is lobbying for its approach and the opportunity to put in place a Googley solution. And what if those outputs are biased. Well, that’s just not possible, is it? One should ask Dr. Timnit Gebru and others who took umbrage at how the estimable Google responded to a bright person’s questions about the broader Google play.
PS. Check out the original research paper, the Snorkel method, and the push back from Xoogler Dr. Gebru. This is an important moment for smart software: To Google or not to Google? That is the question.
Stephen E Arnold, December 7, 2021
Smart Software Is Innovative: Two Marketing Examples, You Doltish Humanoids
December 7, 2021
I zipped through the news releases, headlines, and emails which accumulate in my system. I spotted two stories. Each made the case that smart software — created by humans — is discerning information humans had not previously known or had revealed. This assertion has some interesting implications. There are issues associated with Kurt Gödel-type thinking and the Star Trek think which has launched billions of smart phones.
Here’s the first article. It’s called “AI Is Discovering Patterns in Pure Mathematics That Have Never Been Seen Before.” That’s a clickable title. The write up asserts:
In a newly published study, a research team used artificial intelligence systems developed by DeepMind, the same company that has been deploying AI to solve tricky biology problems and improve the accuracy of weather forecasts, to unknot some long-standing math problems.
DeepMind is pretty much Google. Google is a fan of Snorkel methods. These procedures use minimal training and then let math learn. The outputs are — well — Googley. You know solving the big problems of life like online advertising, reducing the costs of alternative methods of training smart software, and dealing with the legal hassles associated with the alleged “AI cabal” and Timnit Gebru.
The second article is “AI Generates Hypotheses Human Scientists Have Not Thought Of.” The write up says:
One of the benefits of machine learning systems is the way that they can look for patterns and scenarios that programmers didn’t specifically code them to look out for – they take their training data and apply the same principles to new situations. The research shows that this sort of high-speed, ultra-reliable, large-scale data processing can act as an extra tool working with mathematicians’ natural intuition. When you’re dealing with complex, lengthy equations, that can make a significant difference.
What’s interesting is that the write up does not link the researchers with DeepMind. But it appears that the mathematician András Juhász has worked with Googley DeepMind. See “DeepMind AI Collaborates with Humans on Two Mathematical Breakthroughs.”
The first item cited in this blog post appeared on December 4, 2021. The second appeared in October 2021.
My thought is that the Google is injecting rah rah messages about its Snorkel-type approach into highly regarded publications. My hope is that Dr. Timnit Gebru’s and her work gets equal coverage.
Why? The Google wants to be the big dog in certain smart software dog sled pulling. But inbreeding has its downsides; including, bias. PR firms and rah rah marketers are not sensitive to such mathematical oddities as “drift” in my experience. From peer reviewed articles to the open market for “great ideas”, information marches forward on the wheels of propaganda and factual reformation it does, it does.
Stephen E Arnold, December 6, 2021
What Could Possibly Go Wrong: Direct Connections to MSFT SQL Servers?
December 3, 2021
One can now connect Google Data Studio directly to MSSQL servers with a new beta version. Previously, this feat required the use of either Microsoft’s Power BI or Big Query. Reporter Christian Lauer over at CodeX frames this move as an incursion in, “Google Attacks Microsoft Power BI.” He writes:
“Maybe many of you have been waiting for this. Google Data Studio now also offers a connector to MSSQL servers — at least in the beta version. But you can already use it without any problems. For me this is a milestone and a direct attack on Microsoft Power BI. Because now Data Studio is again a bit closer to the top solutions like Power BI and Qlik or Tableau. In addition, you no longer have to use MS products or load the data into a data warehouse beforehand if they come from Microsoft servers. The advantage for Google’s solution is of course that Data Studio is free of charge. … Many companies have MSSQL databases, now the widely used and free Data Studio from Google also offers a built in connector for it. Often the right and better way would be to make the data available via a Data Warehouse or Data Lake. But especially for smaller companies with only a handful of MSSQL databases, this direct way via Data Studio is probably the most efficient.”
The write-up describes the straightforward process for connecting to an MSSQL database via Google Data Studio, complete with a screenshot. For more information, he sends us to Studio’s Help file, “How to Connect to Microsoft SQL Server.” We wonder, though, whether Microsoft would agree this development amounts to an “attack.” The company may barely notice the change. Cyber criminals? We will have to wait and see.
Cynthia Murrell, December 3, 2021
Xooglers Can Define Evil and Want a Judge to Validate Their Definition
December 2, 2021
I read “Google: Former Employees Sue Tech Giant for Allegedly Breaching Don’t Be Evil Pledge.” Nope, not a joke. When I first heard a real live Google spout this phrase to me at a search conference in 1999 in Boston, I thought the shy, perspiring billionaire to be was pulling my leg. I still think that the don’t be evil thing—alleged crafted by Paul Buchheit and Amit Patel — was a high school science club thing. Companies run by anyone but the Googlers had to be evil. The Googlers were a force for good. Right?
Now three employees, assisted by a mini-flock of legal eagles, want to make the company pay big bucks for pitching the don’t be evil line for years. The phrase found its way into assorted company information outputs. I thought I saw it on a Google booth tchotchke shortly after my interaction with the Google billionaire to be in Boston.
How could “real” attorneys, hired by the ultimate science club, use the phrase don’t be evil in corporate outputs? Easy. Lawyers, once housed in trailers, a kilometer from the “real” office were nuisances to be tolerated. The “good” lawyers mostly did what they were supposed to do and rolled with the sci-club.
The write up reports:
The trio had raised concerns at town halls and other forums inside Google about the company potentially selling cloud technology to immigration authorities in the United States, which at the time were engaging in detention tactics considered inhumane by activists.
This appears to be an example of evil.
Perhaps there will be some existential moments in this matter. Google will have to offer an example of being good. Who will decide? A lawyer. Hopefully a member of the high school science club and a person who understands that saying something doesn’t mean anything when money is involved in Silicon Valley.
Stephen E Arnold, December 2, 2021
Right or Wrong to Be Forgotten?
December 2, 2021
While it is still possible to disappear, it is nearly impossible to forget some past mistakes. In 2014, the Court of Justice of the European Union recognized the “right to be forgotten.” The Irish Times reported that Google has something to say on that law, “Google Should Not Get A Say In What Is To Be Forgotten.”
The EU Court of Justice ruled in favor of the “right to be forgotten” against Google’s Spanish subsidiary by Spain’s data protection agency AEPD and a Spanish citizen. The right to be forgotten forces Google to delist information in searches, but the AEPD argued it was in the public’s benefit for information to remain listed.
The biggest issue in question is the current case of the Quinn family against the Irish Bank Resolution Corporation. Should information related to ongoing litigation and national economic concern be removed from the Internet? There is an even bigger question:
“The more fundamental issue which these delistings have drawn attention to, however, is the power of a private company to decide when, and whether, an individual’s right to be forgotten can be enforced. At present, right-to-be-forgotten claims (such as those made in the Quinn case) are considered and decided on by employees of the search-engine operators to whom the request is made. While these search engines publish annual transparency reports which include statistics about how many right-to-be-forgotten applications are made – and how many are successful – these reports do not detail the content of the decisions in right-to-be-forgotten cases – or the factors used in reaching those decisions. The result is that private companies have the power not only to delist articles but to do so based on their own assessment of whether a legitimate right-to-be-forgotten claim exists, what public interest, if any, would require the item to continue appearing in search results, and how to balance any public interest with the data-protection rights of the requesting party.”
There are very few guidelines about how “right to be forgotten” law is applied. Private companies determine who has the right, but how and why do they make that decision?
It sounds like another case of where the present is going to make the standard by which the future will abide.
Whitney Grace, November December 2, 2021
Google Wants to Win at an Epic Game
November 30, 2021
Google is not afraid of other search engines, but it is afraid that Epic Games could topple its already crumbling videogame empire. The Straits Times explores Google’s reaction in the article, “Google Formed ‘Fortnite Task Force’ In Response To Epic Game’s Moves.” Google created the team when Epic released an Android version of Fortnite to Samsung’s Galaxy Store and directly through Epic’s own Web site bypassing the Google Play store.
Epic wants to avoid paying app store commission fees, especially after Google and Apple removed Fortnite from its app stores last year. Epic sued, then Google countersued. Google stated in its counter suit that Epic sold an “unapproved” Fortnite version and it placed users at risk. Epic had their own evidence to share:
“On Monday, Epic responded with a forceful rebuttal of Google’s claims. That filing included details about Google’s “Fortnite Task Force”, which was meant to help cope with the game bypassing its app store. The group met daily in 2018, according to Epic, which cited internal Google documents. As part of its work, the task force latched onto a potential security problem for users installing Fortnite outside of Google’s app store, Epic said.”
Google shared with the media that Epic released an unsafe version. Epic claims that this was a scare tactic, so developers would not release apps outside of Google Play and thus Google would maintain a monopoly on Android apps.
Both Apple and Google have monopolies on their own markets. It sucks, but there is not much way around. Unofficial app releases can gain popularity, but they do run the risk of exposure. Is there any way around this?
Whitney Grace, November 30, 2021
-
- Subscribe to Beyond Search
Feature archive
News archive
-
