After AI Billions, a Hail, Mary Play
November 19, 2024
Now it is scramble time. Reuters reports, “OpenAI and Others Seek New Path to Smarter AI as Current Methods Hit Limitations.” Why does this sound familiar? Perhaps because it is a replay of the enterprise search over-promise and under-deliver approach. Will a new technique save OpenAI and other firms? Writers Krystal Hu and Anna Tong tell us:
“After the release of the viral ChatGPT chatbot two years ago, technology companies, whose valuations have benefited greatly from the AI boom, have publicly maintained that ‘scaling up’ current models through adding more data and computing power will consistently lead to improved AI models. But now, some of the most prominent AI scientists are speaking out on the limitations of this ‘bigger is better’ philosophy. … Behind the scenes, researchers at major AI labs have been running into delays and disappointing outcomes in the race to release a large language model that outperforms OpenAI’s GPT-4 model, which is nearly two years old, according to three sources familiar with private matters.”
One difficulty, of course, is the hugely expensive and time-consuming LLM training runs. Another: it turns out easily accessible data is finite after all. (Maybe they can use AI to generate more data? Nah, that would be silly.) And then there is that pesky hallucination problem. So what will AI firms turn to in an effort to keep this golden goose alive? We learn:
“Researchers are exploring ‘test-time compute,’ a technique that enhances existing AI models during the so-called ‘inference’ phase, or when the model is being used. For example, instead of immediately choosing a single answer, a model could generate and evaluate multiple possibilities in real-time, ultimately choosing the best path forward. This method allows models to dedicate more processing power to challenging tasks like math or coding problems or complex operations that demand human-like reasoning and decision-making.”
OpenAI is using this approach in its new O1 model, while competitors like Anthropic, xAI, and Google DeepMind are reportedly following suit. Researchers claim this technique more closely mimics the way humans think. That couldn’t be just marketing hooey, could it? And even if it isn’t, is this tweak really enough?
Cynthia Murrell, November 19, 2024
AI and Efficiency: What Is the Cost of Change?
November 18, 2024
No smart software. Just a dumb dinobaby. Oh, the art? Yeah, MidJourney.
Companies are embracing smart software. One question which gets from my point of view little attention is, “What is the cost of changing an AI system a year or two down the road?” The focus at this time is getting some AI up and running so an organization can “learn” whether AI works or not. A parallel development is taking place in software vendors enterprise and industry-centric specialized software. Examples range from a brand new AI powered accounting system to Microsoft “sticking” AI into the ASCII editor Notepad.
Thanks, MidJourney. Good enough.
Let’s tally the costs which an organization faces 24 months after flipping the switch in, for example, a hospital chain which uses smart software to convert a physician’s spoken comments about a patient to data which can be used for analysis to provide insight into evidence based treatment for the hospital’s constituencies.
Here are some costs for staff, consultants, and lawyers:
- Paying for the time required to figure out what is on the money and what is not good or just awful like dead patients
- The time required to figure out if the present vendor can fix up the problem or a new vendor’s system must be deployed
- Going through the smart software recompete or rebid process
- Getting the system up and running
- The cost of retraining staff
- Chasing down dependencies like other third party software for the essential “billing process”
- Optimizing the changed or alternative system.
The enthusiasm for smart software makes talking about these future costs fade a little.
I read “AI Makes Tech Debt More Expensive,” and I want to quote one passage from the pretty good essay:
In essence, the goal should be to unblock your AI tools as much as possible. One reliable way to do this is to spend time breaking your system down into cohesive and coherent modules, each interacting through an explicit interface. A useful heuristic for evaluating a set of modules is to use them to explain your core features and data flows in natural language. You should be able to concisely describe current and planned functionality. You might also want to set up visibility and enforcement to make progress toward your desired architecture. A modern development team should work to maintain and evolve a system of well-defined modules which robustly model the needs of their domain. Day-to-day feature work should then be done on top of this foundation with maximum leverage from generative AI tooling.
Will organizations make this shift? Will the hyperbolic AI marketers acknowledge the future costs of pasting smart software on existing software like circus posters on crumbling walls?
Nope.
Those two year costs will be interesting for the bean counters when those kicked cans end up in their workspaces.
Stephen E Arnold, November 18, 2024
Salesforce to MSFT: We Are Coming, Baby Cakes
November 18, 2024
No smart software. Just a dumb dinobaby. Oh, the art? Yeah, MidJourney.
Salesforce, an outfit that hopped on the “attention” bandwagon, is now going whole hog with smart software. “Salesforce to Hire More Than 1,000 Workers to Boost AI Product Sales” makes clear that AI is going to be the hook for the company for the next hype cycle riding toward Next Big Thing theme park.
The write up says:
Agentforce is a new layer on the Salesforce platform, designed to enable companies to build and deploy AI agents that autonomously perform tasks.
Now that’s a buzz packed sentence: “Layer,” sales call data as a “platform”, “AI agents”, “autonomously”, and smart software that can “perform tasks.”
The idea is that sales are an important part of a successful organization. The exception is that monopolies really don’t need too many sales professionals. Lawyers? Yes. Forward deployed engineers? Yes. Marketers? Yes. Door knockers? Well, probably fewer going forward.
How does the Salesforce AI system work? The answer is simple it seems:
These AI agents operate independently, triggered by data changes, business rules, pre-built automations, or API signals.
Who writes the rules? I wonder if AI writes it own rules or do specialists get an opportunity to demonstrate their ability to become essential cogs in the Salesforce customers’ machines?
What do customers do with smart Salesforce? Once again, the answer is easy to provide. The write up says:
Companies such as OpenTable, Saks and Wiley are currently utilizing Agentforce to augment their workforce and enhance customer experiences. Over the past two years, Salesforce has focused on controlling sales expenses by reducing jobs and encouraging customers to use self-service or third-party purchasing options.
I think I understand. Get rid of pesky humans, their vacations, health care, pension plans, and annoying demands for wage increases. Salesforce delivers “efficiency.”
I am not sure what to make of this set of statements. Underpinning Salesforce is a database. The stuff on top of the database are interfaces. Now smart software promises to deliver efficiency and obviously another layer of “smart stuff” to provide what software and services have been promising since the days of the punched card.
Smart software, like Web search, is a natural monopoly unless specific deep pocket outfits can create a defensible niche and sell enough smart software to that niche before some other company eats their lunch.
But that’s what some companies do? Eat other individual’s lunch. So whose taking those lunches tomorrow? Amazon, Google, Microsoft, or Salesforce? Maybe the lunch thief will be a pesky start up essentially off the radar of the big hungry dogs?
With AI development shifting East, is the Silicon Valley AI way the future. Heck, even Google is moving smart software to London which is a heck of a lot easier flight to some innovative locations.
Hopefully one of the AI companies can convert billions in AI investment into new revenue and big profits in a sprightly manner. So far, I see marketing and AI dead ends. Is Salesforce, as the long gone Philco radio company used to say, “The leader”? On one hand, Salesforce is hiring. On the other, get rid of employees. Okay, I think I understand.
Stephen E Arnold, November 18, 2024
Will Tim Apple Vacation in Sochi?
November 18, 2024
![dino-orange_thumb_thumb_thumb_thumb_[1]](https://arnoldit.com/wordpress/wp-content/uploads/2024/11/dino-orange_thumb_thumb_thumb_thumb_1.gif "dino-orange_thumb_thumb_thumb_thumb_[1]")
No smart software. Just a dumb dinobaby. Oh, the art? Yeah, MidJourney.
I love the idea that “It’s just business.” Forget special operations, people falling from windows high above the cobbles, and wheeling and dealing with alleged axis of evil outfits. Focus on doing what is going to sell product. That is the guiding light.
Immanuel Kant, in the midst of pondering his philosophical treatise about ethics, considers the question of the apple on his desk. Thanks, MidJourney. Good enough.
I read the allegedly accurate write up “Apple Removes Another RFE/RL App at Request of Russian Regulator.” The story reports as actual factual:
Roskomnadzor notified Apple that the Russian Service app contains materials from an organization whose activities in Russia have been declared “undesirable.”
What did Apple do (hey, that’s a t-shirt slogan, WDAD)? According to the the cited article:
U.S. technology giant Apple has notified RFE/RL that it has removed another of its apps following a request from Russia’s media regulator, Roskomnadzor. The newly removed RFE/RL app is that of the Russian Service, which in turn hosts the websites of its regional projects Siberia.Realities and North.Realities. Apple had previously removed the apps for RFE/RL’s Kyrgyz Service and Current Time, the Russian-language TV and digital network run by RFE/RL.
The acronym RRE/RL is GenX speak for Radio Free Europe and Radio Liberty. In case you are not familiar with these efforts, the US government funds the broadcasting organizations. The idea is to provide “real” news, information, and analysis (insight) to avid listeners in Eastern Europe, Central Asia, and the Middle East.
The write up adds:
RFE/RL President Stephen Capus called the decision “yet another example of how the Russian government sees truthful reporting as an existential threat.” Besides RFE/RL’s apps, Apple also removed or hid several Russian-language podcasts produced by independent journalists.
From my point of view, the US government wants to provide information to citizens in some countries. Russian authorities do not want that information to flow to residents of those countries. So the Russian authorities told Apple to remove an app which allowed iPhone owners to access certain information deemed unsuitable to citizens in some countries of interest to Russia. I think I am following this … mostly.
Then Apple said, “No problem.” The extremely well-loved Cupertino, California, outfit removed the applications offensive to Russian authorities. Then — let me get this straight in my dinobaby brain — Apple notified Radio Free Europe and Radio Liberty professionals, “Yo, dudes, we rolled over for Vlad and his agents.”
Have I got this right? Apple wants to government to disseminate the information Russia does not like. That’s helpful, Apple.
Several observations:
- Is Apple more powerful in terms of information dissemination than Google and its allegedly reviled video service which has notable performance problems in Russia and parts of the Russian Federation?
- Is the US government supposed to amp up its dissemination of information to the affected nation states? (Well, thanks for the guidance Apple.)
- Is Apple supporting the US government or actively assisting a nation state whose leadership continues to talk about nuclear bombs and existential threats to Mr. Putin’s home base?
My hunch is that Apple, like a handful of other commercial entities, perceives itself as a nation state. The pesky government officials — regardless of where they lives —have to be kept happy. The real objective is keeping those revenues flowing.
Did Immanuel Kant cover this angle in his “Groundwork for the Metaphysics of Morals”? Oh, the apple on Kant’s desk has disintegrated.
Stephen E Arnold, November 18, 2024
E-Casino: Gambling As a Service
November 15, 2024
Gambling is a vice, but it’s also big business. Many gambling practices are illegal and if you want to stay on the right side of the law, then you should make your future gambling business complies with all ordinances. For starters, you need to pay your taxes or the IRS will shut you down. Second, read Revanda Group’s review the “Best White Label Casino Solution Providers In 2024” and see what they offer.
Revpanda Group specializes in iGaming marketing services to assist companies acquire and retain players. They use affiliate marketing strategies to draw and connect traffic with the top brands in their industry. Their entire schtick is helping iGaming companies succeed and stay on the right side of the authorities. Their article is a quick how-to start a casino with the right partners.
Revpanda suggests using a white label casino solution, which is an out-of-the-box solution to start a business:
“…one company provides everything you need, including the casino platform itself, online casino software, payment gateways, an affiliate system, and technical support. Your main responsibilities include creating a logo for the casino website and partnering with an agency for content marketing your brand to potential customers. So, choosing a white label solution is easier than starting your own business from scratch….Simply put, a white label solution provides you with a ready-to-operate casino business whereby a third party will help you maintain and handle everyday operations.”
It almost sounds too good to be true, but Revpanda doesn’t make it sound like a get rich quick scam that are haunting YouTube ads. Revpanda explains that there is upfront cost and risks associate with owning a casino:
“One thing to note is that about 40% revenue share goes to the operator and 60% goes to the platform provider. In essence, white label casino solutions offer a turnkey approach for aspiring casino operators, allowing them to launch and market their business with minimal operational burdens, while sharing revenue with the platform provider.”
The casino-via-Door Dash also recommends potential online gambling parlor operators research their white label casino solution provider recommendations to discover the best fit. They discuss what consider when deciding what provider to work with, including licensing and regulation, game variety and quality, payment solutions, customization options, customer service and support, and mobile compatibility.
Yep, GaaS is a convenience.
Whitney Grace, November 15, 2024
Management Brilliance or Perplexing Behavior
November 15, 2024
Sorry to disappoint you, but this blog post is written by a dumb humanoid. The art? We used MidJourney.
TechCrunch published “Perplexity CEO Offers AI Company’s Services to Replace Striking NYT Staff.” The New York Times Tech Guild went on strike. Aravind Srinivas, formerly at OpenAI and founder of Perplexity, made an interesting offer. According to the cited article, Mr. Srinivas allegedly said he would provide services to “mitigate the effect of a strike by New York Times tech workers.”
A young startup luminary reacts to a book about business etiquette. His view of what’s correct is different from what others have suggested might win friends and influence people. Thanks, MidJourney. Good enough.
Two points: Crossing the picket lines seemed okay if the story is correct and assuming that Perplexity’s smart software would “mitigate the effect” of the strike.
According to the article, “many” people criticized Mr. Srinivas’ offer to help a dead tree with some digital ornaments in a time of turmoil. What the former OpenAI wizard suggested he wanted to do was:
to provide technical infra support on a high traffic day.
Infra, I assume, is infrastructure. And a high-traffic day at a dead tree business is? I just don’t know. The Gray Lady has an online service and it bought an eGame which lacks the bells and whistles of Hamster Kombat. I think that Hamster Kombat has a couple of hundred million users and a revenue stream from assorted addictive elements jazzed with tokens. Could Perplexity help out Telegram if its distributed network ran into more headwinds that the detainment of its founder in France?
Furthermore, the article reminded me that the Top Dog of the dead tree outfit “sent Perplexity a cease and desist letter in October [2024] over the startup’s scraping of articles for use by its AI models.”
What interests me, however, is the outstanding public relations skills that Mr. Srinivas demonstrated. He has captured headlines with his “infra” offer. He is getting traction on Twitter, now the delightfully named X.com. He is teaching old-school executives like Tim Apple how to deal with companies struggling to adapt to the AI, go fast approach to business.
Perplexity’s offer illustrates a conceptual divide between old school publishing, labor unions, and AI companies. Silicon Valley outfits have a deft touch. (I almost typed “tone deaf”. Yikes.)
Stephen E Arnold, November 15, 2024
Microsoft: That Old Time Religion Which Sort of Works
November 15, 2024
Having a favorite OS can be akin to being in a technology cult or following a popular religion. Apple people are experienced enthusiasts, Linux users are the odd ones because it has a secret language and handshakes, while Microsoft is vanilla with diehard followers. Microsoft apparently loves its users and employees to have this mantra and feed into it says Edward Zitron of Where’s Your Ed At? in the article, “The Cult Of Microsoft.”
Zitron reviewed hundreds of Microsoft’s internal documents and spoke with their employees about the company culture. He learned that Microsoft subscribed to “The Growth Mindset” and it determines how far someone will go within the hallowed Redmond halls. There are two types of growth mindset: you can learn and change to continue progressing or you believe everything is immutable (aka the fixed mindset).
Satya Nadella even wrote a bible of sorts called Hit Refresh that discusses The Growth Mindset. Zitron purports that Nadella wants to setup himself up as a messianic figure and used his position to claim a place at the top of the bestseller list. How? He “urged” his Microsoft employees to discuss Hit Refresh with as many people as possible. The communication methods he had his associates use was like a pyramid scheme aka a multi-level marketing ploy.
Microsoft is as fervent of following The Growth Mindset as women used to be selling Mary Kay and Avon products. The problem, Zitron reports, is that it has little to do with actual improvement. The Growth Mindset can’t be replicated without the presence of the original creator.
“In other words, the evidence that supports the efficacy of mindset theory is unreliable, and there’s no proof that this actually improves educational outcomes. To quote Wenner Moyer:
‘MacNamara and her colleagues found in their analysis that when study authors had a financial incentive to report positive effects — because, say, they had written books on the topic or got speaker fees for talks that promoted growth mindset — those studies were more than two and half times as likely to report significant effects compared with studies in which authors had no financial incentives.’
Turning to another view: Wenner Moyer’s piece is a balanced rundown of the chaotic world of mindset theory, counterbalanced with a few studies where there were positive outcomes, and focuses heavily on one of the biggest problems in the field — the fact that most of the research is meta-analyses of other people’s data…”
Microsoft has employees write biannual self-performance reviews called Connects. Everyone hates them but if the employees want raises and to keep their jobs then they have to fill out those forms. What’s even more demeaning is that Copilot is being used to write the Connects. Copilot is throwing out random metrics and achievements that don’t have a basis on any facts.
Is the approach similar to a virtual pyramid scheme. Are employees are taught or hired to externalize their success and internalize their failures. If something the Big Book of MSFT provides grounding in the Redmond way.
Mr. Nadella strikes me as having adopted the principles and mantra of a cult. Will the EU and other regulatory authorities bow before the truth or act out their heresies?
Whitney Grace, November 15, 2024
A Digital Flea Market Tests Smart Software
November 14, 2024
Sales platform eBay has learned some lessons about deploying AI. The company tested three methods and shares its insights in the post, “Cutting Through the Noise: Three Things We’ve Learned About Generative AI and Developer Productivity.” Writer Senthil Padmanabhan explains:
“Through our AI work at eBay, we believe we’ve unlocked three major tracks to developer productivity: utilizing a commercial offering, fine-tuning an existing Large Language Model (LLM), and leveraging an internal network. Each of these tracks requires additional resources to integrate, but it’s not a matter of ranking them ‘good, better, or best.’ Each can be used separately or in any combination, and bring their own benefits and drawbacks.”
The company could have chosen from several existing commercial AI offerings. It settled on GitHub Copilot for its popularity with developers. That and the eBay codebase was already on GitHub. They found the tool boosted productivity and produced mostly accurate documents (70%) and code (60%). The only problem: Copilot’s limited data processing ability makes it impractical for some applications. For now.
To tweak and train an open source LLM, the team chose Code Llama 13B. They trained the camelid on eBay’s codebase and documentation. The resulting tool reduced the time and labor required to perform certain tasks, particularly software upkeep. It could also sidestep a problem for off-the-shelf options: because it can be trained to access data across internal services and within non-dependent libraries, it can get to data the commercial solutions cannot find. Thereby, code duplication can be avoided. Theoretically.
Finally, the team used an Retrieval Augmented Generation to synthesize documentation across disparate sources into one internal knowledge base. Each piece of information entered into systems like Slack, Google Docs, and Wikis automatically received its own vector, which was stored in a vector database. When they queried their internal GPT, it quickly pulled together an answer from all available sources. This reduced the time and frustration of manually searching through multiple systems looking for an answer. Just one little problem: Sometimes the AI’s responses were nonsensical. Were any just plain wrong? Padmanabhan does not say.
The post concludes:
“These three tracks form the backbone for generative AI developer productivity, and they keep a clear view of what they are and how they benefit each project. The way we develop software is changing. More importantly, the gains we realize from generative AI have a cumulative effect on daily work. The boost in developer productivity is at the beginning of an exponential curve, which we often underestimate, as the trouble with exponential growth is that the curve feels flat in the beginning.”
Okay, sure. It is all up from here. Just beware of hallucinations along the way. After all, that is one little detail that still needs to be ironed out.
Cynthia Murrell, November 14, 2024
Pragmatism or the Normalization of Good Enough
November 14, 2024
Sorry to disappoint you, but this blog post is written by a dumb humanoid. The art? We used MidJourney.
I recall that some teacher told me that the Mona Lisa painter fooled around more with his paintings than he did with his assistants. True or false? I don’t know. I do know that when I wandered into the Louvre in late 2024, there were people emulating sardines. These individuals wanted a glimpse of good old Mona.
Is Hamster Kombat the 2024 incarnation of the Mona Lisa? I think this image is part of the Telegram eGame’s advertising. Nice art. Definitely a keeper for the swipers of the future.
I read “Methodology Is Bullsh&t: Principles for Product Velocity.” The main idea, in my opinion, is do stuff fast and adapt. I think this is similar to the go-go mentality of whatever genius said, “Go fast. Break things.” This version of the Truth says:
All else being equal, there’s usually a trade-off between speed and quality. For the most part, doing something faster usually requires a bit of compromise. There’s a corner getting cut somewhere. But all else need not be equal. We can often eliminate requirements … and just do less stuff. With sufficiently limited scope, it’s usually feasible to build something quickly and to a high standard of quality. Most companies assign requirements, assert a deadline, and treat quality as an output. We tend to do the opposite. Given a standard of quality, what can we ship in 60 days? Recent escapades notwithstanding, Elon Musk has a similar thought process here. Before anything else, an engineer should make the requirements less dumb.
Would the approach work for the Mona Lisa dude or for Albert Einstein? I think Al fumbled along for years, asking people to help with certain mathy issues, and worrying about how he saw a moving train relative to one parked at the station.
I think the idea in the essay is the 2024 view of a practical way to get a product or service before prospects. The benefits of redefining “fast” in terms of a specification trimmed to the MVP or minimum viable product makes sense to TikTok scrollers and venture partners trying to find a pony to ride at a crowded kids’ party.
One of the touchstones in the essay, in my opinion, is this statements:
Our customers are engineers, so we generally expect that our engineers can handle product, design, and all the rest. We don’t need to have a whole committee weighing in. We just make things and see whether people like them.
I urge you to read the complete original essay.
Several observations:
- Some people like the Mona List dude are engaged in a process of discovery, not shipping something good enough. Discovery takes some people time, lots of time. What happens during this process is part of the process of expanding an information base.
- The go-go approach has interesting consequences; for example, based on the anecdotal and flawed survey data, young users of social media evidence a number of interesting behaviors. The idea of “let ‘er rip” appears to have some impact on young people. Perhaps you have one hand experience with this problem? I know people whose children have manifested quite remarkable behaviors. I do know that certain basic mental functions like concentrating is visible to me every time I have a teenager check me out at the grocery store.
- By redefining excellence and quality, the notion of a high-value goal drops down a bit. Some new automobiles don’t work too well; for example, the Tesla Cybertruck owner whose vehicle was not able to leave the dealer’s lot.
Net net: Is a Telegram mini app Hamster Kombat today’s equivalent of the Mona Lisa?
Stephen E Arnold, November 14, 2024
Marketers, Deep Fakes Work
November 14, 2024
Bad actors use AI for pranks all the time, but this could be the first time AI pranked an entire Irish town out of its own volition. KXAN reports on the folly: “AI Slop Site Sends Thousands In Ireland To Fake Halloween Parade.” The website MySpiritHalloween.com dubs itself as the ultimate resource for all things Halloween. The website uses AI-generated content and an article told the entire city of Dublin that there would be a parade.
If this was a small Irish village, there would giggles and the police would investigate criminal mischief before they stopped wasting their time. Dublin, however, is one of the country’s biggest cities and folks appeared in the thousands to see the Halloween parade. They eventually figured something was wrong without the presence of barriers, law enforcement, and (most importantly) costume people on floats!
MySpiritHalloween is owned by Naszir Ali and he was embarrassed by the situation.
Per Ali’s explanation, his SEO agency creates websites and ranks them on Google. He says the company hired content writers who were in charge of adding and removing events all across the globe as they learned whether or not they were happening. He said the Dublin event went unreported as fake and that the website quickly corrected the listing to show it had been cancelled….
Ali said that his website was built and helped along by the use of AI but that the technology only accounts for 10-20% of the website’s content. He added that, according to him, AI content won’t completely help a website get ranked on Google’s first page and that the reason so many people saw MySpiritHalloween.com was because it was ranked on Google’s first page — due to what he calls “80% involvement” from actual humans.”
Ali claims his website is based in Illinois, but all investigations found that its hosted in Pakistan. Ali’s website is one among millions that use AI-generated content to manipulate Google’s algorithm. Ali is correct that real humans did make the parade rise to the top of Google’s search results, but he was responsible for the content.
Media around the globe took this as an opportunity to teach the parade goers and others about being aware of AI-generated scams. Marketers, launch your AI infused fakery.
Whitney Grace, November 14, 2024
-
- Subscribe to Beyond Search
Feature archive
News archive
-
