Instragram Does the YouTube Creator Fear Thing
November 11, 2024
Instagram influencers are enraged by CEO Adam Mosseri’s bias towards popular content. According to the BBC in, “Instagram Lowering Quality Of Less Viewed Videos ‘Alarming’ Creators”, video quality is lowered for older, less popular videos. More popular content gets the HD white glove treatment. Influencers are upset over this “discrimination,” especially when they concentrate on making income through Instagram over other platforms.
The influences view the lower quality output as harmful and affects the quality of original art. Mosseri argues that most influencers have their videos watched soon after publication. The only videos being affected by lower quality are older and no longer receive many views. While that sounds logical, it could also create a cycle that benefits only a few influencers:
Social media consultant Matt Navarra told the BBC the move ‘seems to somewhat contradict Instagram’s earlier messages or efforts to encourage new creators’.
"How can creators gain traction if their content is penalized for not being popular," he said. And he said it could risk creating a cycle of more established creators reaping the rewards of higher engagement from viewers over those trying to build their following.”
Instagram is lowering the quality to save on costs. It always comes down money, doesn’t it? When asked to respond about that, Mosseri said viewers are more interested in a video’s content over its image quality. Navarra agreed to that statement:
“He [Navarra] said creators should focus on how they can make engaging content that caters to their audience, rather than be overly concerned by the possibility of its quality being degraded by Instagram.”
Navarra’s right. Video quality will be decent and not poor like a cathode-ray tube TV. The creators should focus on building themselves and not investing all of their creative energy into one platform. Diversify!
Whitney Grace, November 11, 2024
FOGINT: Crypto Is a Community Builder
November 9, 2024
CreationNetwork.ai: A One-Stop Shop for All Your Digital Needs. And Crypto, Too!
Here is an interesting development. “CreationNetwork.ai Emerges As a Leading AI-Powered Platform, Integrating Over Twenty Two Tools,” reports HackerNoon. The AI aggregator uses Telegram plus other social media to push its service. Furthermore, the company is integrating crypto into its business plan. We expect these “blending” plays will become more common. The Chainwire press release says about this one:
“As an all-in-one solution for content creation, e-commerce, social media management, and digital marketing, CreationNetwork.ai combines 22+ proprietary AI-powered tools and 29+ platform integrations to deliver the most extensive digital ecosystem available. … CreationNetwork.ai’s suite of tools spans every facet of digital engagement, equipping users with powerful AI technologies to streamline operations, engage audiences, and optimize performance. Each tool is meticulously designed to enhance productivity and efficiency, making it easy to create, manage, and analyze content across multiple channels.”
See the write-up for a list of the tools included in CreationNetwork.ai, from AI Copywriter to Team-Powered Branding. The hefty roster of platform connections is also specified, including obvious players: all the major social media platforms, the biggest e-commerce platforms, and content creation tools like Canva, Grammarly, Adobe Express, Unsplash, and Dropbox. We learn:
“One of the most distinguishing features of CreationNetwork.ai is its extensive integration network. With over 29 integrations, users can synchronize their digital activities across major social media, e-commerce, and content platforms, providing centralized management and engagement capabilities. … This integration network empowers users to manage their brand presence across platforms from a single, unified dashboard, significantly enhancing efficiency and reach.”
Nifty. What a way to simplify digital processes for users. And to make it harder for new services to break into the market. But what groundbreaking platform would be complete without its own cryptocurrency? The write-up states:
“In preparation for its Initial Coin Offering (ICO), CreationNetwork.ai is launching a $750,000 CRNT Token Airdrop to reward early supporters and incentivize participation in the CreationNetwork.ai ecosystem. Qualified participants can secure their position by following CreationNetwork.ai’s social media accounts and completing the whitelist form available on the official website. This initiative highlights CreationNetwork.ai’s commitment to building a strong, engaged community.”
Crypto — The community builder.
Cynthia Murrell, November 11, 2024
Meta and China: Yeah, Unauthorized Use of Llama. Meh
November 8, 2024
This post is the work of a dinobaby. If there is art, accept the reality of our using smart art generators. We view it as a form of amusement.
That open source smart software, you remember, makes everything computer- and information-centric so much better. One open source champion laboring as a marketer told me, “Open source means no more contractual handcuffs, the ability to make changes without a hassle, and evidence of the community.
An AI-powered robot enters a meeting. One savvy executive asks in Chinese, “How are you? Are you here to kill the enemy?” Another executive, seated closer to the gas emitted from a cannister marked with hazardous materials warnings gasps, “I can’t breathe!” Thanks, Midjourney. Good enough.
How did those assertions work for China? If I can believe the “trusted” outputs of the “real” news outfit Reuters, just super cool. “Exclusive: Chinese Researchers Develop AI Model for Military Use on Back of Meta’s Llama”, those engaging folk of the Middle Kingdom:
… have used Meta’s publicly available Llama model to develop an AI tool for potential military applications, according to three academic papers and analysts.
Now that’s community!
The write up wobbles through some words about the alleged Chinese efforts and adds:
Meta has embraced the open release of many of its AI models, including Llama. It imposes restrictions on their use, including a requirement that services with more than 700 million users seek a license from the company. Its terms also prohibit use of the models for “military, warfare, nuclear industries or applications, espionage” and other activities subject to U.S. defense export controls, as well as for the development of weapons and content intended to “incite and promote violence”. However, because Meta’s models are public, the company has limited ways of enforcing those provisions.
In the spirit of such comments as “Senator, thank you for that question,” a Meta (aka Facebook), wizard allegedly said:
“That’s a drop in the ocean compared to most of these models (that) are trained with trillions of tokens so … it really makes me question what do they actually achieve here in terms of different capabilities,” said Joelle Pineau, a vice president of AI Research at Meta and a professor of computer science at McGill University in Canada.
My interpretation of the insight? Hey, that’s okay.
As readers of this blog know, I am not too keen on making certain information public. Unlike some outfits’ essays, Beyond Search tries to address topics without providing information of a sensitive nature. For example, search and retrieval is a hard problem. Big whoop.
But posting what I would term sensitive information as usable software for anyone to download and use strikes me as something which must be considered in a larger context; for example, a bad actor downloading an allegedly harmless penetration testing utility of the Metasploit-ilk. Could a bad actor use these types of software to compromise a commercial or government system? The answer is, “Duh, absolutely.”
Meta’s founder of the super helpful Facebook wants to bring people together. Community. Kumbaya. Sharing.
That has been the lubricant for amassing power, fame, and money… Oh, also a big gold necklace similar to the one’s I saw labeled “Pharaoh jewelry.”
Observations:
- Meta (Facebook) does open source for one reason: To blunt initiatives from its perceived competitors and to position itself to make money.
- Users of Meta’s properties are only data inputters and action points; that is, they are instrumentals.
- Bad actors love that open source software. They download it. They study it. They repurpose it to help the bad actors achieve their goals.
Did Meta include a kill switch in its open source software? Oh, sure. Meta is far-sighted, concerned with misuse of its innovations, and super duper worried about what an adversary of the US might do with that technology. On the bright side, if negotiations are required, the head of Meta (Facebook) allegedly speaks Chinese. Is that a benefit? He could talk with the weaponized robot dispensing biological warfare agents.
Stephen E Arnold, November 8, 2024
Microsoft Does the Me Too with AI. Si, Wee Wee
November 8, 2024
Sorry to disappoint you, but this blog post is written by a dumb humanoid. The art? We used MidJourney.
True or false: “Windows Intelligence” or Microsoft dorkiness? The article “Windows Intelligence: Microsoft May Drop Copilot in Major AI Rebranding” contains an interesting subtitle. Here it is:
Copying Apple for marketing’s sake, as usual.
Is that a snarky subtitle or not? It follows up the idea of “rebranding” Satya Nadella’s next big thing with an assertion of possibly Copy Cat behavior. I like it.
The article says:
Copilot’s chatbot service, Windows Recall, and other AI-related functionalities could soon become known simply as “Windows Intelligence.” According to a recently surfaced reference included in a template file for the Group Policy Object Editor (AppPrivacy.adml), Windows Intelligence is the umbrella term Microsoft possibly chose to unify the many AI features currently being integrated into the PC operating system.
Whether such an attempt to capture the cleverness of Apple intelligence (AI for short) with the snappy WI is a stroke of genius or a clumsy me too move, time will reveal.
Let’s assume that the Softies take time out from their labors on the security issues bedeviling users and apply its considerable expertise to the task of move from Copilot to WI. One question which will arise may be, “How does one pronounce WI?”
I personally like “whee” as in the sound seven year olds make when sliding an old fashioned playground toy. I can envision some people using the WI sound in the word “whiff.” The sound of the first two letters of “weak” is a possibility as well. And, one could emit the sound “wuh” which rhymes with “Duh.”
My hunch is that this article’s assertion is unsubstantiated beyond a snippet of text in “code.” Nevertheless I enjoyed the write up. My hunch is that the author had some fun putting the article together as well.
Oh, Copilot image generation has been down for more than a weak, oh, sorry, week. That’s very close to the whiff sound for the alleged WI acronym. Oh, well, a whiff by any other name should reveal so much about a high technology giant.
Stephen E Arnold, November 8, 2024
Penalty for AI Generated Child Abuse Images
November 8, 2024
Whenever new technology is released it’s only a matter of time before a bad actor uses it for devious purposes. Those purposes are usually a form of sex, theft, and abuse. Bad actors saw a golden opportunity with AI image generation for child pornography and ArsTechnica reported that: “18-Year Prison Sentence For Man Who Used AI To Create Child Abuse Images.” Hugh Nelson, the pedophile from the UK used a 3D AI software to make child sexual abuse imagery. When his crime was discovered, he was sentences to eighteen years in prison. It’s a landmark case for prosecuting deepfakes in the UK.
Nelson used Daz 3D to make the sexually explicit images. AI image algorithms use large data models to generate “new” images. The algorithms can also take preexisting images and alter them. Nelson used photographs of real children, fed them into Daz 3D, and had deepfake SA images. He also encouraged other bad actors to do the same thing. Nelson will be incarcerated until he completes two-thirds of his sentence. The judge at the trial said Nelson was a “significant risk” to the public.
Since these images are fake, one could argue that they’re harmless but the problem here was the use of real children’s images. These real kids had their visage transformed into sexually explicit images. That’s where the debate about harm and intent enters:
“Graeme Biggar, director-general of the UK’s National Crime Agency, last year warned it had begun seeing hyper-realistic images and videos of child sexual abuse generated by AI. He added that viewing this kind of material, whether real or computer-generated, “materially increases the risk of offenders moving on to sexually abusing children themselves.”
Greater Manchester Police’s specialist online child abuse investigation team said computer-generated images had become a common feature of their investigations.
‘This case has been a real test of the legislation, as using computer programs in this particular way is so new to this type of offending and isn’t specifically mentioned within current UK law,’ detective constable Carly Baines said when Nelson pleaded guilty in August. The UK’s Online Safety Act, which passed last October, makes it illegal to disseminate non-consensual pornographic deepfakes. But Nelson was prosecuted under existing child abuse law.”
My personal view is that Nelson should be locked up for the remainder of his putrid existence as should the people who asked him to make those horrible images. Don’t mess with kids!
Whitney Grace, November 8, 2024
Let Them Eat Cake or Unplug: The AI Big Tech Bro Effect
November 7, 2024
I spotted a news item which will zip right by some people. The “real” news outfit owned by the lovable Jeff Bezos published “As Data Centers for AI Strain the Power Grid, Bills Rise for Everyday Customers.” The write up tries to explain that AI costs for electric power are being passed along to regular folks. Most of these electricity dependent people do not take home paychecks with tens of millions of dollars like the Nadellas, the Zuckerbergs, or the Pichais type breadwinners do. Heck, these AI poohbahs think about buying modular nuclear power plants. (I want to point out that these do not exist and may not for many years.)
The article is not going to thrill the professionals who are experts on utility demand and pricing. Those folks know that the smart software poohbahs have royally screwed up some weekends and vacations for the foreseeable future.
The WaPo article (presumably blessed by St. Jeffrey) says:
The facilities’ extraordinary demand for electricity to power and cool computers inside can drive up the price local utilities pay for energy and require significant improvements to electric grid transmission systems. As a result, costs have already begun going up for customers — or are about to in the near future, according to utility planning documents and energy industry analysts. Some regulators are concerned that the tech companies aren’t paying their fair share, while leaving customers from homeowners to small businesses on the hook.
Okay, typical “real” journospeak. “Costs have already begun going up for customers.” Hey, no kidding. The big AI parade began with the January 2023 announcement that the Softies were going whole hog on AI. The lovable Google immediately flipped into alert mode. I can visualize flashing yellow LEDs and faux red stop lights blinking in the gray corridors in Shoreline Drive facilities if there are people in those offices again. Yeah, ghostly blinking.
The write up points out, rather unsurprisingly:
The tech firms and several of the power companies serving them strongly deny they are burdening others. They say higher utility bills are paying for overdue improvements to the power grid that benefit all customers.
Who wants PEPCO and VEPCO to kill their service? Actually, no one. Imagine life in NoVa, DC, and the ever lovely Maryland without power. Yikes.
From my point of view, informed by some exposure to the utility sector at a nuclear consulting firm and then at a blue chip consulting outfit, here’s the scoop.
The demand planning done with rigor by US utilities took a hit each time the Big Dogs of AI brought more specialized, power hungry servers online and — here’s the killer, folks — and left them on. The way power consumption used to work is that during the day, consumer usage would fall and business/industry usage would rise. The power hogging steel industry was a 24×7 outfit. But over the last 40 years, manufacturing has wound down and consumer demand crept upwards. The curves had to be plotted and the demand projected, but, in general, life was not too crazy for the US power generation industry. Sure, there were the costs associated with decommissioning “old” nuclear plants and expanding new non-nuclear facilities with expensive but management environmental gewgaws, gadgets, and gizmos plugged in to save the snail darters and the frogs.
Since January 2023, demand has been curving upwards. Power generation outfits don’t want to miss out on revenue. Therefore, some utilities have worked out what I would call sweetheart deals for electricity for AI-centric data centers. Some of these puppies suck more power in a day than a dying city located in Flyover Country in Illinois.
Plus, these data centers are not enough. Each quarter the big AI dogs explain that more billions will be pumped into AI data centers. Keep in mind: These puppies run 24×7. The AI wolves have worked out discount rates.
What do the US power utilities do? First, the models have to be reworked. Second, the relationships to trade, buy, or “borrow” power have to be refined. Third, capacity has to be added. Fourth, the utility rate people create a consumer pricing graph which may look like this:
Guess who will pay? Yep, consumers.
The red line is the prediction for post-AI electricity demand. For comparison, the blue line shows the demand curve before Microsoft ignited the AI wars. Note that the gray line is consumer cost or the monthly electricity bill for Bob and Mary Normcore. The nuclear purple line shows what is and will continue to happen to consumer electricity costs. The red line is the projected power demand for the AI big dogs.
The graph shows that the cost will be passed to consumers. Why? The sweetheart deals to get the Big Dog power generation contracts means guaranteed cash flow and a hurdle for a low-ball utility to lumber over. Utilities like power generation are not the Neon Deions of American business.
There will be hand waving by regulators. Some city government types will argue, “We need the data centers.” Podcasts and posts on social media will sprout like weeds in an untended field.
Net net: Bob and Mary Normcore may have to decide between food and electricity. AI is wonderful, right.
Stephen E Arnold, November 7, 2024
Google: The Intellectual Oakland Because There Is No There There, Just Ads
November 7, 2024
The post is the work of a humanoid who happens to be a dinobaby. GenX, Y, and Z, read at your own risk. If art is included, smart software produces these banal images.
I read a clever essay titled “I Attended Google’s Creator Conversation Event, And It Turned Into A Funeral.” Be aware that the text can disappear as a big gray box covers the essay. Just read quickly.
The report explains that a small group of problematic content creators found their sites or other content effectively made invisible in Google search results. Boom. Videos disappear. Boom. Key words no longer retrieve a Web site. So Google did the politically correct thing and got a former conference organizer to round up some of the disaffected and come to a meeting to discuss content getting disappeared. No real reasons are given by Google, but the essay recounts the experience on one intrepid optimist who thought, “This time Google will be different.”
A very professional Googler evades a question. A person in the small group meeting asks it again. She is not happy with Google’s evasiveness. Well, too bad. Thanks, Midjourney. MSFT Copilot is still dead.
Nope.
The write up does a good job of capturing the nothingness of a Google office. If you have not been to one, try to set up a meeting. Good luck.
I want to focus on a couple of points in the essay and then offer a handful of observations.
I noted this statement:
During this small group discussion, I and others tried to get our Googlers to address the biggest problem facing our industry: Google giving big brands special treatment. Each time a site owner brought up the topic, we were quickly steered in another direction.
Google wants or assumes that it is control of everything most of the time. Losing control means one is not a true Googler. The people at this Google event learned that non-Googlers and their questions are simply not relevant. Google goes through certain theatrical events to check off a task.
Also, I circled this comment:
…we then asked the only question that mattered: Why has Google shadow banned our sites? Google’s Chief Search Scientist answered this question using a strategy based around gaslighting and said they hadn’t. Google doesn’t ever derank an entire site, only individual pages, he said. There is no site-wide classifier. He insisted it is only done at the page level.
This statement is accurate. A politically correct answer is one that does not reveal Google’s intent for anything. Individuals in charge of a project or program usually do not know what is going on with that program. Information is shared via the in house communications systems, email or text messages which are viewed as potential problems if released to those outside the Google, and meetings which are often ad hoc or without an agenda. The company sells advertising and reacts to what appear to be threats, legal actions, or ways to make money.
I noted this statement:
someone bluntly asked, since nothing is wrong with our sites, how do we recover? Google’s elderly Chief Search Scientist answered, without an ounce of pity or concern, that there would be updates but he didn’t know when they’d happen or what they’d do. Further questions on the subject were met with indifference as if he didn’t understand why we cared. He’d gotten the information he wanted. The conference was over. I don’t think he even said thanks.
This is accurate. Why should a member of leadership care about a “user”? Clicks produce data. The data and attendant content are processed to make more money. That’s why customer service is limited to big budget advertisers who spend real money on Google advertising.
Several observations:
- Google is big (150,000 or more employees). Google is chaotic. Individuals, groups, and entire divisions are not sure what is going on. How many messaging apps did Google have at one time? Lots. Why? No one knows or knew. The people coming to the meeting about finding themselves invisible in the Google finding systems assumed that a big company was not chaotic. Now the attendees know.
- People who create content have replaced people who used to get paid to create content. Now the “creators” work for the hope of Google advertising money. What’s the percentage paid to a “creator”? Try to find those data, gentle reader. Google does what it does: Individual Googlers or a couple of Googlers set up a system and go to play Foosball. That means 149,998 colleagues have zero idea what’s happening. Content “creators” expect someone to be responsible and to know how systems work. The author of the essay now knows only a couple of people may know the answer to the question. If those people quit, one will never know.
- The people who use Google to find relevant information are irrelevant as individuals. The person who wants to find a pizza in Cleveland may find only the pizza a person working on Google Local for Cleveland may like. If that pizza joint spends a couple of thousand per month on Google ads, that pizza may be findable. Most people do not understand the linkages between search engine optimization, Google advertising sales, and the Google mostly automated Google ad auctioning system. One search engineer working from home can have quite an impact on people who make content and assume that Google will make it findable. The author of the article knows this assumption about Google is a fairy tale.
Google has be labeled a monopoly. Google is suggesting that if the company is held accountable for its behaviors, the US will lose its foothold in artificial intelligence. Brilliant argument. Google has employees who have won a Nobel Prize. People not held accountable often lose sight of many things.
That’s why the big meeting was dumped into the task list of a person who ran search engine optimization conferences. One does not pray for that individual; one does not go to meetings managed by such a professional.
Stephen E Arnold, November 7, 2024
The Yogi Berra Principle: Déjà Vu All Over Again
November 7, 2024
Sorry to disappoint you, but this blog post is written by a dumb humanoid. The art? We used MidJourney.
I noted two write ups. The articles share what I call a meta concept. The first article is from PC World called “Office Apps Crash on Windows 11 24H2 PCs with CrowdStrike Antivirus.” The actors in this soap opera are the confident Microsoft and the elegant CrowdStrike. The write up says:
The annoying error affects Office applications such as Word or Excel, which crash and become completely unusable after updating to Windows 11 24H2. And this apparently only happens on systems that are running antivirus software by CrowdStrike. (Yes, the very same CrowdStrike that caused the global IT meltdown back in July.)
A patient, skilled teacher explains to the smart software, “You goofed, speedy.” Thanks, Midjourney. Good enough.
The other write up adds some color to the trivial issue. “Windows 11 24H2 Misery Continues, As Microsoft’s Buggy Update Is Now Breaking Printers – Especially on Copilot+ PCs” says:
Neowin reports that there are quite a number of complaints from those with printers who have upgraded to Windows 11 24H2 and are finding their device is no longer working. This is affecting all the best-known printer manufacturers, the likes of Brother, Canon, HP and so forth. The issue is mainly being experienced by those with a Copilot+ PC powered by an Arm processor, as mentioned, and it either completely derails the printer, leaving it non-functional, or breaks certain features. In other cases, Windows 11 users can’t install the printer driver.
Okay, dinobaby, what’s the meta concept thing? Let me offer my ideas about the challenge these two write ups capture:
- Microsoft may have what I call the Boeing disease; that is, the engineering is not so hot. Many things create problems. Finger pointing and fast talking do not solve the problems.
- The entities involved are companies and software which have managed to become punch lines for some humorists. For instance, what software can produce more bricks than a kiln? Answer: A Windows update. Ho ho ho. Funny until one cannot print a Word document for a Type A, drooling MBA.
- Remediating processes don’t remediate. The word process itself generates flawed outputs. Stated another way, like some bank mainframes and 1960s code, fixing is not possible and there are insufficient people and money to get the repair done.
The meta concept is that the way well paid workers tackle an engineering project is capable of outputting a product or service that has a failure rate approaching 100 percent. How about those Windows updates? Are they the digital equivalent of the Boeing space initiative.
The answer is, “No.” We have instances of processes which cannot produce reliable products and services. The framework itself produces failure. This idea has some interesting implications. If software cannot allow a user to print, what else won’t perform as expected? Maybe AI?
Stephen E Arnold, November 7, 2024
Dreaming about Enterprise Search: Hope Springs Eternal…
November 6, 2024
The post is the work of a humanoid who happens to be a dinobaby. GenX, Y, and Z, read at your own risk. If art is included, smart software produces these banal images.
Enterprise search is back, baby. The marketing lingo is very year 2003, however. The jargon has been updated, but the story is the same: We can make an organization’s information accessible. Instead of Autonomy’s Neurolinguistic Programming, we have AI. Instead of “just text,” we have video content processed. Instead of filters, we have access to cloud-stored data.
An executive knows he can crack the problem of finding information instantly. The problem is doing it so that the time and cost of data clean up does not cost more than buying the Empire State Building. Thanks, Stable Diffusion. Good enough.
A good example of the current approach to selling the utility of an enterprise search and retrieval system is the article / interview in Betanews called “How AI Is Set to Democratize Information.” I want to be upfront. I am a mostly aligned with the analysis of information and knowledge presented by Taichi Sakaiya. His The Knowledge Value Revolution or a History of the Future has been a useful work for me since the early 1990s. I was in Osaka, Japan, lecturing at the Kansai Institute of Technology when I learned of this work book from my gracious hosts and the Managing Director of Kinokuniya (my sponsor). Devaluing knowledge by regressing to the fat part of a Gaussian distribution is not something about which I am excited.
However, the senior manager of Pyron (Raleigh, North Carolina), an AI-powered information retrieval company, finds the concept in line with what his firm’s technology provides to its customers. The article includes this statement:
The concept of AI as a ‘knowledge cloud’ is directly tied to information access and organizational intelligence. It’s essentially an interconnected network of systems of records forming a centralized repository of insights and lessons learned, accessible to individuals and organizations.
The benefit is, according to the Pyron executive:
By breaking down barriers to knowledge, the AI knowledge cloud could eliminate the need for specialized expertise to interpret complex information, providing instant access to a wide range of topics and fields.
The article introduces a fresh spin on the problems of information in organizations:
Knowledge friction is a pervasive issue in modern enterprises, stemming from the lack of an accessible and unified source of information. Historically, organizations have never had a singular repository for all their knowledge and data, akin to libraries in academic or civic communities. Instead, enterprise knowledge is scattered across numerous platforms and systems — each managed by different vendors, operating in silos.
Pyron opened its doors in 2017. After seven years, the company is presenting a vision of what access to enterprise information could, would, and probably should do.
The reality, based on my experience, is different. I am not talking about Pyron now. I am discussing the re-emergence of enterprise search as the killer application for bolting artificial intelligence to information retrieval. If you are in love with AI systems from oligopolists, you may want to stop scanning this blog post. I do not want to be responsible for a stroke or an esophageal spasm. Here we go:
- Silos of information are an emergent phenomenon. Knowledge has value. Few want to make their information available without some value returning to them. Therefore, one can talk about breaking silos and democratization, but those silos will be erected and protected. Secret skunk works, mislabeled projects, and squirreling away knowledge nuggets for a winter’s day. In the case of Senator Everett Dirksen, the information was used to get certain items prioritized. That’s why there is a building named after him.
- The “value” of information or knowledge depends on another person’s need. A database which contains the antidote to save a child from a household poisoning costs money to access. Why? Desperate people will pay. The “information wants to free” idea is not one that makes sense to those with information and the knowledge to derive value from what another finds inscrutable. I am not sure that “democratizing information” meshes smoothly with my view.
- Enterprise search, with or without, hits some cost and time problems with a small number of what have been problems for more than 50 years. SMART failed, STAIRS III failed, and the hundreds of followers have failed. Content is messy. The idea that one can process text, spreadsheets, Word files, and email is one thing. Doing it without skipping wonky files or the time and cost of repurposing data remains difficult. Chemical companies deal with formulae; nuclear engineering firms deal with records management and mathematics; and consulting companies deal with highly paid people who lock up their information on a personal laptop. Without these little puddles of information, the “answer” or the “search output” will not be just a hallucination. The answer may be dead wrong.
I understand the need to whip up jargon like “democratize information”, “knowledge friction”, and “RAG frameworks”. The problem is that despite the words, delivering accurate, verifiable, timely on-point search results in response to a query is a difficult problem.
Maybe one of the monopolies will crack the problem. But most of output is a glimpse of what may be coming in the future. When will the future arrive? Probably when the next PR or marketing write up about search appears. As I have said numerous times, I find it more difficult to locate the information I need than at any time in my more than half a century in online information retrieval.
What’s easy is recycling marketing literature from companies who were far better at describing a “to be” system, not a “here and now” system.
Stephen E Arnold, November 4, 2024
Twenty Five Percent of How Much, Google?
November 6, 2024
The post is the work of a humanoid who happens to be a dinobaby. GenX, Y, and Z, read at your own risk. If art is included, smart software produces these banal images.
I read the encomia to Google’s quarterly report. In a nutshell, everything is coming up roses even the hyperbole. One news hook which has snagged some “real” news professionals is that “more than a quarter of new code at Google is generated by AI.” The exclamation point is implicit. Google’s AI PR is different from some other firms; for example, Samsung blames its financial performance disappointments on some AI. Winners and losers in a game in which some think the oligopolies are automatic winners.
An AI believer sees the future which is arriving “soon, real soon.” Thanks, You.com. Good enough because I don’t have the energy to work around your guard rails.
The question is, “How much code and technical debt does Google have after a quarter century of its court-described monopolistic behavior? Oh, that number is unknown. How many current Google engineers fool around with that legacy code? Oh, that number is unknown and probably for very good reasons. The old crowd of wizards has been hit with retirement, cashing in and cashing out, and “leadership” nervous about fiddling with some processes that are “good enough.” But 25 years. No worries.
The big news is that 25 percent of “new” code is written by smart software and then checked by the current and wizardly professionals. How much “new” code is written each year for the last three years? What percentage of the total Google code base is “new” in the years between 2021 and 2024? My hunch is that “new” is relative. I also surmise that smart software doing 25 percent of the work is one of those PR and Wall Street targeted assertions specifically designed to make the Google stock go up. And it worked.
However, I noted this Washington Post article: “Meet the Super Users Who Tap AI to Get Ahead at Work.” Buried in that write up which ran the mostly rah rah AI “real” news article coincident with Google’s AI spinning quarterly reports one interesting comment:
Adoption of AI at work is still relatively nascent. About 67 percent of workers say they never use AI for their jobs compared to 4 percent who say they use it daily, according to a recent survey by Gallup.
One can interpret this as saying, “Imagine the growth that is coming from reduced costs. Get rid of most coders and just use Google’s and other firms’ smart programming tools.
Another interpretation is, “The actual use is much less robust than the AI hyperbole machine suggests.”
Which is it?
Several observations:
- Many people want AI to pump some life into the economic fuel tank. By golly, AI is going to be the next big thing. I agree, but I think the Gallup data indicates that the go go view is like looking at a field of corn from a crop duster zipping along at 1,000 feet. The perspective from the airplane is different from the person walking amidst the stalks.
- The lack of data behind Google-type assertions about how much machine code is in the Google mix sounds good, but where are the data? Google, aren’t you data driven? So, where’s the back up data for the 25 percent assertion.
- Smart software seems to be something that is expensive, requires dreams of small nuclear reactors next to a data center adjacent a hospital. Yeah, maybe once the impact statements, the nuclear waste, and the skilled worker issues have been addressed. Soon as measured in environmental impact statement time which is different from quarterly report time.
Net net: Google desperately wants to be the winner in smart software. The company is suggesting that if it were broken apart by crazed government officials, smart software would die. Insert the exclamation mark. Maybe two or three. That’s unlikely. The blurring of “as is” with “to be” is interesting and misleading.
Stephen E Arnold, November 6, 2024
-
- Subscribe to Beyond Search
Feature archive
News archive
-
