THCB

Categories

Tag: Kim Bellard

America the Schizophrenic

By KIM BELLARD

I must admit, last week’s election took me by surprise. I knew all the polls predicted a close race, but I kept telling myself that the American I believed in would not elect such a man, again, knowing full well all the things he has said and done – in his personal, professional, and political lives.  I was giving us too much credit.

Democrats might tell the public that Wall Street was hitting record highs, that GDP growth was among the best in the world, that unemployment was low, and that inflation was finally back under control, but voters didn’t believe them. For most people, the economy isn’t working.

When two-thirds of voters say the country is on the wrong track (NBC News), when almost three-quarters of Americans are dissatisfied with the way things are going in the U.S. (Gallop), when 62% of voters think the economy is weak and 48% say their personal financial situation is getting worse (Harvard CAPS/Harris) – well, threats to democracy tomorrow don’t compare to the price of eggs today.  

Let’s face it: we are on the wrong road. We’re not on a road that is good for most people. We’re not on a road that is getting us ready for the challenges and opportunities that the 21st century is bringing/is going to bring us. And we’re kidding ourselves about the America we believe in versus the America we actually live in.  Our views about our country are delusional, they’re disorganized thinking, they may even be hallucinations. I.e., they’re schizophrenic. 

For example:

Continue reading…

Engineers: Heal Thyselves (and Health Care)

By KIM BELLARD

The article I can’t get out of my head is one by Greg Ip in The Wall Street Journal: Crises at Boeing and Intel Area National Emergency.

I’m old enough that I remember when the Boeing 707 took airline passenger travel from the prop age to the jet age. I’m old enough that I remember that we all wanted PCs with Intel chips when companies starting giving office workers their first PCs. I’ve read enough history to know the storied engineering background and achievements of both. I mean, those B-52s that have been the backbone of the U.S. Air Force bomber command for the past 70+ years: those are Boeing planes.

To younger people, though, Being is the company whose doors pop out mid-flight, or which abandons astronauts in space. When they think of Intel – oh, I’m just kidding; when younger people think about chip companies, it’s NVIDIA or TSMC. Intel’s stock is doing so badly it may get kicked out of the Dow Jones Industrial Average.

So, as Mr. Ip says: “A generation ago, any list of America’s most admired manufacturers would have had Intel and Boeing near the top. Today, both are on the ropes.”

He goes on to add:

The U.S. still designs the world’s most innovative products, but is losing the knack for making them.

At the end of 1999, four of the 10 most valuable U.S. companies were manufacturers. Today, none are. The lone rising star: Tesla, which ranked 11th.

Intel and Boeing were once the gold standard in manufacturing groundbreaking products to demanding specifications with consistently high quality. Not any longer. 

What is most frustrating, Mr. Ip points out, is: “Neither fell prey to cheap foreign competition, but to their own mistakes. Their culture evolved to prioritize financial performance over engineering excellence.”

As an example, in a Blockbuster-could-have-bought-Netflix parallel, The New York Times reports that Intel could have bought NVIDIA in 2005, but the reported $20b price was considered too expensive. NVIDIA is now worth $3.5 trillion. Whoops.

Boeing’s new CEO, Kelly Ortberg, admits: “The trust in our company has eroded,” and that Boeing needs “a fundamental change in culture.” It doesn’t help that its machinists have been on strike almost 2 months, with the union rejecting Boeing’s latest offer last week. Boeing is slashing some 17,000 jobs, considering selling off its Starliner business, and trying to raise as much as $25b. 

Intel has also cut jobs, is trying to beef up its manufacturing through a revitalized foundry business (which some believe Intel should spin off), and has seen its stock crater (down 52% YTD), but CEO Pat Gelsinger vows: “We see the finish line in sight.”

Intel is still waiting for some $8.5b in CHIPS Act funding, “There’s been renegotiations on both sides,” Mr. Gelsinger told The New York Times. “My simple message is, ‘Let’s get it finished.’” But, as former Commerce Department official Caitlin Legacki noted: [There is fear that] Intel is going to take chips money, build an empty shell of a factory and then never actually open it, because they don’t have customers.”  Its much-hyped plants in Arizona and Ohio have both faced setbacks. 

Meanwhile, the vultures are circling: there are rumors that Samsung and Apple may want to acquire Intel.

The trouble is, which is Mr. Ip’s point, neither has any real domestic competition; if either would fail, it would throw even more of our economy to the mercy of foreign manufacturers (or, in its space business, make the U.S. even more dependent on Elon Musk’s SpaceX). That’s the national emergence he is warning about.

My point with all this is not so much to add another lament about the decline of U.S. manufacturing as to emphasize the decline of the role of engineers. Earlier this year Jerry Useem, writing in The Atlantic,  argued: “When the wave of Japanese competition finally crashed on corporate America, those best equipped to understand it—the engineers—were no longer in charge. American boardrooms had been handed over to the finance people.”   

 Mr. Useem points out that a revitalized GE “is belatedly yielding to the reality that workers on the gemba [Japanese term for the shop floor, where value is actually created] are far better at figuring out more efficient ways of making things than remote bureaucrats with spreadsheet abstractions.” That sounds a lot like what Mr. Ortberg is saying: “We need to be on the factory floors, in the back shops and in our engineering labs.”

So what, you might ask, does this have to do with healthcare? 

It turns out that there is something called a healthcare engineer.

Continue reading…

Health Care’s Endless Loops

By KIM BELLARD

Last week the Federal Trade Commission (FTC) issued its final “click-to-cancel” rule, making it easier for consumers to cancel various kinds of subscriptions, such as gym memberships or streaming services. It will require enrollments to be as easy to cancel as they were to enroll.

“Too often, businesses make people jump through endless hoops just to cancel a subscription,” said Commission Chair Lina M. Khan. “The FTC’s rule will end these tricks and traps, saving Americans time and money. Nobody should be stuck paying for a service they no longer want.”

Oh, boy, Chairperson Khan: if you want to talk about jumping through endless loops, let’s talk about health care.

The FTC rule was part of its effort to modernize its 1973 Negative Option Rule. It had issued a preliminary rule in March 2023, which drew some 16,000 comments. Laura Brett, vice president of the National Advertising Division of BBB National Programs, explained the need for the rule to CNN: “(Consumers) had to jump through hoops online to find out where to cancel. Other times they might’ve been able to sign up online, but in order to cancel they had to call and talk to a representative. Other kinds of memberships required them to actually show up in person to cancel their subscription,”

The new rule is also part of a broader Biden Administration Time Is Money initiative, “a new governmentwide effort to crack down on all the ways that corporations—through excessive paperwork, hold times, and general aggravation—add unnecessary headaches and hassles to people’s days and degrade their quality of life.”

Predictably, not everyone agrees. The U.S. Chamber of Commerce called the whole Time is Money initiative a heavy-handed effort to micromanage business practices and pricing, and warned it would lead to “fewer choices, higher prices, and more headaches.”

But of course they do; I mean, if you asked an AI to create a Chamber of Commerce response to virtually any regulation, it would probably sound much like that.

Critics see politics behind the rule. In her dissent, Melissa Holyoak, one of the FTC’s two Republican commissioners, wrote: “Why the rush? There is a simple explanation. Less than a month from election day, the Chair is hurrying to finish a rule that follows through on a campaign pledge made by the Chair’s favored presidential candidate.” The same could be said of the Biden Administration’s new proposed rules to make over-the-counter birth control to be covered by insurers at no cost to patients.

Be that as it may, we all have more subscriptions than we probably really want, the gym membership scam has been going on so long that there was a Friends episode about it almost 30 years ago, and who among us hasn’t gotten caught in endless loops with supposed customer service representatives – if you can ever reach a live person – about some problem with a company?

Which leads me to health care.

Providing health care has always been complex, as physicians like to remind us, but just trying to receive health care has grown more and more complex over the past several decades (while growing drastically more expensive). Time is Money, the Biden Administration tells us, but in health care, the only people whose time is valued are the people billing us. We are, after all, patients, so we are supposed to be patient.

The FTC, and the Biden Administration more generally, has this right: Time is Money, and that’s our time and our money. The initiative specifically included healthcare – “…the unnecessary complications of dealing with health insurance companies…” – but I don’t think that goes far enough, fast enough.

I like the precept that things should be as easy to get out of as they were to get into, although I want to use that more as a parameter than a restriction in expanding the discussion to healthcare.

Continue reading…

You’re Not Going to Automate MY Job

By KIM BELLARD

Earlier this month U.S. dockworkers struck, for the first time in decades. Their union, the International Longshoremen’s Association (ILW), was demanding a 77% pay increase, rejecting an offer of a 50% pay increase from the shipping companies. People worried about the impact on the economy, how it might impact the upcoming election, even if Christmas would be ruined. Some panic hoarding ensued.

Then, just three days later, the strike was over, with an agreement for a 60% wage increase over six years. Work resumed. Everyone’s happy right? Well, no. The agreement is only a truce until January 15, 2025. While money was certainly an issue – it always is – the real issue is automation, and the two sides are far apart on that.

Most of us aren’t dockworkers, of course, but their union’s attitude towards automation has lessons for our jobs nonetheless.

The advent of shipping containers in the 1960’s (if you haven’t read The Box: How the Shipping Container Made the World Smaller and the World Economy Bigger, by Marc Levinson, I highly recommend it) made increased use of automation in the shipping industry not only possible but inevitable. The ports, the shipping companies, and the unions all knew this, and have been fighting about it ever since. Add better robots and, now, AI to the mix, and one wonders when the whole process will be automated.

Curiously, the U.S. is not a leader in this automation. Margaret Kidd, program director and associate professor of supply chain logistics at the University of Houston, told The Hill: “What most Americans don’t realize is that American exceptionalism does not exist in our port system. Our infrastructure is antiquated. Our use of automation and technology is antiquated.”

Eric Boehm of Reason agrees:

The problem is that American ports need more automation just to catch up with what’s considered normal in the rest of the world. For example, automated cranes in use at the port of Rotterdam in the Netherlands since the 1990s are 80 percent faster than the human-operated cranes used at the port in Oakland, California, according to an estimate by one trade publication.

The top rated U.S. port in the World Bank’s annual performance index is only 53rd.  

Continue reading…

Red Alert about Red Buttons

By KIM BELLARD

In a week where, say, the iconic brand Tupperware declared bankruptcy and University of Michigan researchers unveiled a squid-inspired screen that doesn’t use electronics, the most startling stories have been about, of all things, pagers and walkie-talkies.

Now, most of us don’t think much about either pagers or walkie-talkies these days, and when we do, we definitely don’t think about them exploding. But that’s what happened in Lebanon this week, in ones carried by members of Hezbollah. Scores of people were killed and thousands injured, many of them innocent bystanders. The suspicion, not officially confirmed, is that Israel engineered the explosions.

I don’t want to get into a discussion about the Middle East quagmire, and I condemn the killing of innocent civilians on either side, but what I can’t get my mind around is the tradecraft of the whole thing. This was not a casual weekend cyberattack by some guys sitting in their basements; this was a years-in-the-making, deeply embedded, carefully planned move.

A former Israeli intelligence official told WaPo that, first, intelligence agencies had to determine “what Hezbollah needs, what are its gaps, which shell companies it works with, where they are, who are the contacts,” then “you need to create an infrastructure of companies, in which one sells to another who sells to another.”  It’s not clear, for example, if Israel someone planted the devices during the manufacturing process or during the shipping, or, indeed, if its shell companies actually were the manufacturer or shipping company. 

Either way, this is some James Bond kind of shit.

The Washington Post reports that this is what Israeli officials call a “red-button” capability, “meaning a potentially devastating penetration of an adversary that can remain dormant for months if not years before being activated.” One has to wonder what other red buttons are out there.

Many have attributed the attacks to Israel’s Unit 8200, which is roughly equivalent to the NSA.  An article in Reuters described the unit as “famous for a work culture that emphasizes out-of-the-box thinking to tackle issues previously not encountered or imagined.”  Making pagers explode upon command certainly falls in that category.

If you’re thinking, well, I don’t carry either a pager or a walkie-talkie, and, in any event, I’m not a member of Hezbollah, don’t be so quick to think you are off the hook. If you use a device that is connected to the internet – be it a phone, a TV, a car, even a toaster – you might want to be wondering if it comes with a red button. And who might be in control of that button.

Just today, for example, the Biden Administration proposed a ban on Chinese software used in cars.

Continue reading…

We Should Learn to Have More Fun (or Vice-Versa)

By KIM BELLARD

For several years now, my North Star for thinking about innovation has been Steven Johnson’s great quote (in his delightful Wonderland: How Play Made the Modern World): “You will find the future where people are having the most fun.” No, no, no, naysayers argue, inventing the future is serious business, and certainly fun is not the point of business.  Maybe they’re right, but I’m happier hoping for a future guided by a sense of fun than by one guided by P&Ls.

Well, I think I may have found an equally insightful point of view about fun, espoused by game designer Raph Koster in his 2004 book A Theory of Fun for Game Design: “Fun is just another word for learning.”

Wow.

That’s not how most of us think about learning. Learning is hard, learning is going to school, learning is taking tests, learning is something you have to do when you’re not having fun. So “fun is just another word for learning” is quite a different perspective – and one I’m very much attracted to.

I regret that it took me twenty years to discover Mr. Koster’s insight. I read it in a more current book: Kelly Clancy’s Playing With Reality: How Games Have Shaped Our World. Dr. Clancy is not a game designer; she is a neuroscientist and physicist, but she is all about play. Her book looks at games and game theory, especially how the latter has been misunderstood/misused.

We usually think of play as a waste of time, as something inherently unserious and unimportant, when, in fact, it is how our brains have evolved to learn. The problem is, we’ve turned learning into education, education into a requirement, teaching into a profession, and fun into something entirely separate. We’ve gotten it backwards.

“Play is a tool the brain uses to generate data on which to train itself, a way of building better models of the world to make better predictions,” she writes. “Games are more than an invention; they are an instinct.”  Indeed, she asserts: “Play is to intelligence as mutation is to evolution.”

Mr. Koster’s fuller quote about fun and learning is on target with this:

That’s what games are, in the end. Teachers. Fun is just another word for learning. Games teach you how aspects of reality work, how to understand yourself, how to understand the actions of others, and how to imagine.

We don’t look at our teachers as a source of fun (and many students barely look at them as a source of learning). We don’t look at schools as a place for games, except on the playground, and then only for the youngest students. We drive students to boredom, and, as Mr. Koster says, “boredom is the opposite of learning” (although, ironically, boredom may be important to creativity).  

Learning is actually fun, especially from a physiological standpoint.

Continue reading…

The Fantastic Fungi — Biohybrid Bots Are Mushrooming

By KIM BELLARD

I hadn’t expected to write about a biology-related topic anytime soon after doing so last week, but, gosh darn it, then I saw a press release from Cornell about biohybrid robots – powered by mushrooms (aka fungi)! They had me at “biohybrid.”  

The release talks about a new paper — Sensorimotor Control of Robots Mediated by Electrophysiological Measurements of Fungal Mycelia – from the Cornell’s Organic Robotics Lab, led by Professor Rob Shepherd. As the release describes the work:

By harnessing mycelia’s innate electrical signals, the researchers discovered a new way of controlling “biohybrid” robots that can potentially react to their environment better than their purely synthetic counterparts.  

Or, in the researchers’ own words:

The paper highlights two key innovations: first, a vibration- and electromagnetic interference–shielded mycelium electrical interface that allows for stable, long-term electrophysiological bioelectric recordings during untethered, mobile operation; second, a control architecture for robots inspired by neural central pattern generators, incorporating rhythmic patterns of positive and negative spikes from the living mycelia.

Let’s simplify that: “This paper is the first of many that will use the fungal kingdom to provide environmental sensing and command signals to robots to improve their levels of autonomy,” Professor Shepherd said. “By growing mycelium into the electronics of a robot, we were able to allow the biohybrid machine to sense and respond to the environment.”

Lead author Anand Mishra, a research associate in the lab, explained: “If you think about a synthetic system – let’s say, any passive sensor – we just use it for one purpose. But living systems respond to touch, they respond to light, they respond to heat, they respond to even some unknowns, like signals. That’s why we think, OK, if you wanted to build future robots, how can they work in an unexpected environment? We can leverage these living systems, and any unknown input comes in, the robot will respond to that.”

The team build two robots: a soft one shaped like a spider, and a wheeled one. The researchers first used the natural spike in the mycelia to make them walk and roll, respectively, using the natural signals from the mycelia. Then researchers exposed them to ultraviolet light, which caused the mycelia to react and changed the robots’ gaits. Finally, the researchers were able to override the mycelia signals entirely.

“This kind of project is not just about controlling a robot,” Dr. Mishra said. “It is also about creating a true connection with the living system. Because once you hear the signal, you also understand what’s going on. Maybe that signal is coming from some kind of stresses. So you’re seeing the physical response, because those signals we can’t visualize, but the robot is making a visualization.”

Dr. Shepherd believes that instead of using light as the signal, they will use chemical signals. For example: “The potential for future robots could be to sense soil chemistry in row crops and decide when to add more fertilizer, for example, perhaps mitigating downstream effects of agriculture like harmful algal blooms.”

It turns out that biohybrid robots in general and fungal computing in particular are a thing. In last week’s article I quoted Professor Andrew Adamatzky, of the University of the West of England about his preference for fungal computing. He not only is the Professor in Unconventional Computing there, and is the founder and Editor-in-Chief of the International Journal for Unconventional Computing, but also literally wrote the book about fungal computing.  He’s been working on fungal computing since 2018 (and before that on slime mold computing).

Professor Adamatzky notes that fungi have a wide array of sensory inputs: “They sense light, chemicals, gases, gravity, and electric fields,” which opens the door to a wide variety of inputs (and outputs). Accordingly, Ugnius Bajarunas, a member of Professor Adamatzy’s team, told an audience last year: “Our goal is real-time dialog between natural and artificial systems.”

With fungal computing, TechHQ predicts: “The future of computing could turn out to be one where we care for our devices in a way that’s closer to looking after a houseplant than it is to plugging in and switching on a laptop.”

But how would we reboot them?

Continue reading…

Biology to the Rescue?

By KIM BELLARD

I feel much about synthetic biology as I do AI: I don’t really understand it from a technical point of view, but I sure am excited about its potential. Sometimes they even overlap, as I’ll discuss later. But I’ll start with some recent developments with bioplastics, a topic I have somehow never really covered.

Let’s start with some work at Washington University (St. Louis) involving, of all things, purple bacteria. In case you didn’t know it – I certainly didn’t – purple bacteria “are a special group of aquatic microbes renowned for their adaptability and ability to create useful compounds from simple ingredients,” according to the press release. The researchers are turning the bacteria into bioplastic factories.

One study, led by graduate student Eric Connors, showed that two “obscure” species of purple bacteria can produce polyhydroxyalkanoates (PHAs), a natural polymer that can be purified to make plastics.  Another study, led by research lab supervisor Tahina Ranaivoarisoa, took another “well studied but notoriously stubborn” species of purple bacteria to dramatically ramp up its production of PHAs, by inserting a gene that helped turn them into “relative PHA powerhouses.” The researchers are optimistic they could use other bacteria to produce even higher levels of bioplastics.

The work was done in the lab of associate professor Aripta Bose, who said: “There’s a huge global demand for bioplastics. They can be produced without adding CO2 to the atmosphere and are completely biodegradable. These two studies show the importance of taking multiple approaches to finding new ways to produce this valuable material.”

“It’s worth taking a look at bacteria that we haven’t looked at before,” Mr. Conners said. “We haven’t come close to realizing their potential.” Professor Bose agrees: “We hope these bioplastics will produce real solutions down the road.”

Meanwhile, researchers at Korea Advanced Institute of Science and Technology, led by Sang Yup Lee, have manipulated bacteria to produce polymers that contain “ring-like structures,” which apparently make the plastics more rigid and thermally stable.  Normally those structures would be toxic to the bacteria, but the researchers managed to enable E. coli bacteria to both tolerate and produce them.  The researchers believe that the polymer would be especially useful in biomedical applications, such as drug delivery.

As with the Washington University work, this research is not producing output at scale, but the researchers have good confidence that it can. “If we put more effort into increasing the yield, then this method might be able to be commercialized at a larger scale,” says Professor Lee. “We’re working to improve the efficiency of our production process as well as the recovery process, so that we can economically purify the polymers we produce.”

Because the polymer is produced using biological instead of chemical processes, and is biodegradable, the researchers believe it can be important for the environment. “I think biomanufacturing will be a key to the success of mitigating climate change and the global plastic crisis,” says Professor Lee. “We need to collaborate internationally to promote bio-based manufacturing so that we can ensure a better environment for our future.”

Environmental impact is also very much on the minds of researchers at the University of Virginia. They are working on creating biodegradable bioplastics from food waste. “By creating cost-effective bioplastics that naturally decompose, we can reduce plastic pollution on land and in oceans and address significant issues such as greenhouse gas emissions and economic losses associated with food waste,” said lead researcher Zhiwu “Drew” Wang.

The team is developing microorganisms that convert food waste into fats, which are then processed into bioplastics. Those bioplastics then should easily be composed. “Our first step is to make single-layer film to see if it can be utilized as an actual product,” said Chenxi Cao, a senior in packaging and system design. “If it has good oxygen and water vapor barriers and other properties, we can move to the next step. We aim to replace traditional coated paper products with PHA. Current paper products are often coated with polyethylene or polyactic acid, which are not fully degradable. PHA is fully biodegradable in nature, even in a backyard environment.”

The approach is currently still in the pilot project stage.

If all that isn’t cool enough, our own bodies may become biofactories, such as to deliver drugs or vaccines. Earlier this year researchers at UT Southwestern reported on “in situ production and secretion of proteins,” which in this case targeted psoriasis and two types of cancer.

The researchers say: “Through this engineering approach, the body can be utilized as a bioreactor to produce and systemically secrete virtually any encodable protein that would otherwise be confined to the intracellular space of the transfected cell, thus opening up new therapeutic opportunities.”

“Instead of going to the hospital or outpatient clinic frequently for infusions, this technology may someday allow a patient to receive a treatment at a pharmacy or even at home once a month, which would be a significant boost to their quality of life,” said study leader Daniel Siegwart, Ph.D. Professor Siegwart believes this type of in situ production could eventually improve health and quality of life for patients with inflammatory diseases, cancers, clotting disorders, diabetes, and a range of genetic disorders.  

I promised I’d touch on an example of synthetic biology and AI overlapping. Last year I wrote about how “organoid intelligence” was a new approach to biocomputing and AI. Earlier this year Swiss firm FinalSpark launched its Neuroplatform, which uses 16 human brain organoids as the computing platform, claiming it was: “The next evolutionary leap for AI.”   

“Our principal goal is artificial intelligence for 100,000 times less energy,” FinalSpark co-founder Fred Jordan says. 

Now FinalSpark is renting its biocomputers to AI researchers at several top universities…for only $500 a month. “As far as I know, we are the only ones in the world doing this” on a publicly rentable platform, Dr. Jordan told Scientific American. Reportedly, around 34 universities requested access, but FinalSpark so far has limited use to 9 institutions, including the University of Michigan, the Free University of Berlin, and the Lancaster University in Germany.

Scientific America reports related work at Spain’s National Center for Biotechnology, using cellular computing, and at the University of the West of England, using – I’m serious! – fungal networks. “Fungal computing offers several advantages over brain-organoid-based computing,” Andrew Adamatzky says, “particularly in terms of ethical simplicity, ease of cultivation, environmental resilience, cost-effectiveness and integration with existing technologies.”

Bioplastics, biofactories, biocomputing — pretty cool stuff all around. I’ll admit I don’t know where all of this is leading, but I can’t wait to see where it leads.   

Tiny Is Mighty

By KIM BELLARD

I am a fanboy for AI; I don’t really understand the technical aspects, but I sure am excited about its potential. I’m also a sucker for a catchy phrase. So when I (belatedly) learned about TinyAI, I was hooked.  

Now, as it turns out, TinyAI (also know as Tiny AI) has been around for a few years, but with the general surge of interest in AI it is now getting more attention. There is also TinyML and Edge AI, the distinctions between which I won’t attempt to parse. The point is, AI doesn’t have to involve huge datasets run on massive servers somewhere in the cloud; it can happen on about as small a device as you care to imagine. And that’s pretty exciting.

What caught my eye was a overview in Cell by Farid Nakhle, a professor at Temple University, Japan Campus: Shrinking the Giants: Paving the Way for TinyAI.  “Transitioning from the landscape of large artificial intelligence (AI) models to the realm of edge computing, which finds its niche in pocket-sized devices, heralds a remarkable evolution in technological capabilities,” Professor Nakhle begins.

AI’s many successes, he believes, “…are demanding a leap in its capabilities, calling for a paradigm shift in the research landscape, from centralized cloud computing architectures to decentralized and edge-centric frameworks, where data can be processed on edge devices near to where they are being generated.” The demands for real time processing, reduced latency, and enhanced privacy make TinyAI attractive.

Accordingly: “This necessitates TinyAI, here defined as the compression and acceleration of existing AI models or the design of novel, small, yet effective AI architectures and the development of dedicated AI-accelerating hardware to seamlessly ensure their efficient deployment and operation on edge devices.”

Professor Nakhle gives an overview of those compression and acceleration techniques, as well as architecture and hardware designs, all of which I’ll leave as an exercise for the interested reader.  

If all this sounds futuristic, here are some current examples of TinyAI models:

  • This summer Google launched Gemma 2 2B, a 2 billion parameter model that it claims outperforms OpenAI’s GPT 3.5 and Mistral AI’s Mixtral 8X7B. VentureBeat opined: “Gemma 2 2B’s success suggests that sophisticated training techniques, efficient architectures, and high-quality datasets can compensate for raw parameter count.”
  • Also this summer OpenAI introduced GPT-4o mini, “our most cost-efficient small model.” It “supports text and vision in the API, with support for text, image, video and audio inputs and outputs coming in the future.”
  • Salesforce recently introduced its xLAM-1B model, which it likes to call the “Tiny Giant.” It supposedly only has 1b parameters, yet Marc Benoff claims it outperforms modelx 7x its size and boldly says: “On-device agentic AI is here”  
  • This spring Microsoft launched Phi-3 Mini, a 3.8 billion parameter model, which is small enough for a smartphone. It claims to compare well to GPT 3.5 as well as Meta’s Llama 3.
  • H2O.ai offers Danube 2, a 1.8 b parameter model that Alan Simon of Hackernoon calls the most accurate of the open source, tiny LLM models.   

A few billion parameters may not sound so “tiny,” but keep in mind that other AI models may have trillions.

Continue reading…

It’s in the Blood

By KIM BELLARD

People are fascinated by blood. Well, it would seem so, given our fondness for vampires, gory movies, and true crime stories. I’m not so keen on any of those, but I was struck by several recent developments about how blood tests can help diagnose medical problems faster, more definitively, and less invasively.

Because, really, shouldn’t that be what our healthcare system always should strive for?

Take concussions. If you are a football fan, you’re very familiar with the problem that it seems very subjective about whether a player has suffered a concussion. They’re not the only ones. Millions of people suffer concussions each year – the vast majority of whom are not athletes – and more than half never get it evaluated.

In April Abbott received FDA approval for a rapid blood test, producing results in 15 minutes. It can be done at a patient’s beside, and not require a lab. “Clinicians have needed an objective way to assess patients with concussions,” said Beth McQuiston, M.D., medical director in Abbott’s diagnostics business. “When you look at all the other diseases, or other organs in the body, they all have blood tests to help assess what’s happening. Now, we have a whole blood test that can help assess the brain right at the patient’s bedside – expanding access to more health providers and therefore patients.”

Expect to see the Abbott’s i-STAT TBI cartridge and portable i-STAT® Alinity® instrument in emergency rooms, not to mention on NFL sidelines.

Or Alzheimer’s disease. Many realize that it has historically been very difficult to diagnose, often not definitively until after death. Now a new study suggests a blood test can accurately diagnose it 90% of the time, which is much higher than even neurologists can do. The test is more accurate the later the stage of Alzheimer’s a person has.

Specifically, it measures “the ratio of plasma phosphorylated tau 217 (p-tau217) relative to non–p-tau217 (expressed as percentage of p-tau217) combined with the amyloid-β 42 and amyloid-β 40 plasma ratio (the amyloid probability score 2 [APS2]).” Got that?

“We’d love to have a blood test that can beused in a primary care physician’s office, functioning like a cholesterol test but for Alzheimer’s,” Dr. Maria Carrillo, chief science officer of the Alzheimer’s Association, told CNN. “The p-tau217 blood test is turning out to be the most specific for Alzheimer’s and the one with the most validity. It seems to be the front-runner.”

It’s not quite ready for use in your doctor’s office, though. “Right now, we don’t have guidelines for the use of these tests,” Dr. Eliezer Masliah, director of the division of neuroscience at the National Institute on Aging, warned NPR. Dr. Suzanne Schindler, an associate professor of neurology at Washington University School of Medicine in St. Louis, added: “Blood tests have developed incredibly fast for Alzheimer disease and I think [doctors] aren’t used to that rate of change.”

“The field is moving at a pace I never imagined 10 years ago,” Dr. Heather Whitson, a professor of medicine at Duke University, marveled to NPR.

We’re increasingly seeing FDA approved drug treatments for Alzheimer’s, so it’d be nice if we had FDA approved blood tests to more accurately use them. 

Last but not least, there’s colorectal cancer (CRC). The FDA recently approved Guardant Health’s Shield™ blood test for colorectal cancer screening, and is the first such blood test approved by the FDA as a primary screening option. A Guardant study found that it identified 87% of cancers that were at an early and curable stage, although it does less well at finding precancerous growths. The test is aimed at adults 45 and older who are at average risk.

It’s not so much that it is better than colonoscopies – it’s not — as it is that it should be easier to convince people to use. Despite the fact that CRC kills over 50,000 Americans annually, more than a third of older Americans are not getting screened. Even worse, more than three-fourths of those who die from CRC are not up-to-date with their screening.

“The persistent gap in colorectal cancer screening rates shows that the existing screening options do not appeal to millions of people,” said Daniel Chung, MD, gastroenterologist at Massachusetts General Hospital and Professor of Medicine at Harvard Medical School. “The FDA’s approval of the Shield blood test marks a tremendous leap forward, offering a compelling new solution to close this gap.”

Dr. Sapna Syngal, director of strategic planning for prevention and early cancer detection at the Dana-Farber Cancer Center in Boston agrees, telling NBC News: “If this test increases the number of people getting screened, it’s going to have a huge impact.” 

The test is on the market now, and Guardant expects approval for coverage by Medicare and commercial insurance.

Most of us are used to getting routine blood tests for things like blood counts or cholesterol levels, so it’s exciting that blood tests are started to be used for other important health issues.

————–

Blood tests are all well and good, but they’re not (yet) the kind of test you’d routinely expect to use at home on your own. ARPA-H has even bigger aspirations. It just announced the Platform Optimizing SynBio for Early Intervention and Detection in Oncology (POSEIDON) program, the goal of which is “to develop first-in-class, at-home, synthetic Multi-Cancer-Early Detection (MCED) tests for the most sensitive and specific stage I detection of 30+ solid tumors* using only breath and/or urine samples.”

No blood draw or lab tech needed, just breath or urine samples done yourself at home. That’s something to shoot for.

“Access to a low-cost cancer screening test that does not need a lab test is so critical to preventing late-stage diagnoses, increasing survival rates, and reducing high treatment costs,” said ARPA-H Director Renee Wegrzyn, Ph.D. “With POSEIDON, we could put the power of cancer screening into homes in the U.S. and around the world.” 

“But what if any adult could, at their discretion, take an at-home test that could detect Stage I cancer? POSEIDON aims to create a future in which any adult can take a simple, over-the-counter test to screen for and detect 30+ cancers at Stage I, when they are still localized, to drastically improve the chances of curative treatment and survival,” said Ross Uhrich, DMD, MBA, ARPA-H POSEIDON’s Program Manager.

“But what if…” indeed.  ARPA-H is thinking big — as it should. And as should we all.

Kim is a former emarketing exec at a major Blues plan, editor of the late & lamented Tincture.io, and now regular THCB contributor