IBM Eats its Own AI Dog Food

Granted, you’d expect a technology firm to tout the benefits of using the tools it sells. Consider IBM, which highlights its positioning as a hybrid cloud supplier.  “Across IBM's IT environment, we're realizing the value of hybrid cloud,” said James Kavanaugh, IBM CFO. “We reduced the average cost of running an application by 90 percent by moving from a legacy data center environment to a hybrid cloud environment running on Red Hat OpenShift.” IBM, of course, owns Red Hat. 

“By standardizing global processes and applying AIOps, we are reducing our application portfolio by more than 35 percent,” he adds. “We've automated over 24 million transactions with RPA (robotic process automation), avoiding hundreds of thousands of manual tasks and eliminating the risk of human error.”

IBM also is applying artificial intelligence. “in HR, we now handle 94 percent of our company-wide HR inquiries with our AskHR digital system, speeding up the completion of many HR tasks by up to 75 percent.”

Large Language Models will Save Chatbots

Chatbots almost always suck. But large language models and generative AI should vastly improve that experience. Yeah, we know: almost anything would fix something that is so unsatisfying. But generative AI will allow answers to a far-greater range of questions and an improved way of fixing matters automatically. 

Not only does generative AI allow indexing a wider range of possible information, it also aids code writing routines. And better code generation should also assist automated responses (and fixes) for stated customer issues. 

Large Language Model Inflection Point?

For most people, it seems as though artificial intelligence has suddenly emerged as an idea and set of possibilities. In truth, AI has been gestating for many many decades. But forms of AI already are used in consumer appliances such as smart speakers, recommendation engines and search functions.

What seems to be happening now is some inflection point in adoption. But consider the explosion of interest in large language models or generative AI. Development has been under way for more than 70 years. 

Search engines, smart phones and smart speakers have been using AI to support speech interfaces. In that sense, consumers have routinely been using AI-assisted devices and apps for some time. 

Large Language Models, or Generative AI, have lots of potential applications in virtually any setting where language, questions and answers or content creation--including development of computer code--is involved. 

source: Java T Point

Compared to earlier supervised learning models, large language models are self-supervised, able to scour huge amounts of internet data to predict the next word in a sentence. 

A large language model “is a type of artificial intelligence (AI) algorithm that uses deep learning techniques and massively large data sets to understand, summarize, generate and predict new content,” consultant Sean Kerner says. 

Right now, the obvious use cases are text summarization, chatbots, search, and code generation. Other use cases undoubtedly will develop. That suggests early use for customer service, text generation, writing of code and information retrieval tasks. 

It seems clear that large language models, as a subset of AI, have reached an inflection point. What remains unclear is the degree of progress and adoption. At most inflection points there is a quantitative shift in usage that often leads to qualitative impact. 

We will see quantitative change a lot faster than potential qualitative effects, near term. The qualitative changes will take longer, but should be far deeper than we now envision. That is just the way technology change tends to happen.

Generative AI Progress: Less than You Expect, Near Term; More than You Imagine Long Term

For most people, it seems as though artificial intelligence has suddenly emerged as an idea and set of possibilities. In truth, AI has been gestating for many many decades. But forms of AI already are used in consumer appliances such as smart speakers, recommendation engines and search functions.

What seems to be happening now is some inflection point in adoption. But consider the explosion of interest in large language models or generative AI. Development has been under way for more than 70 years. 

source: Black Hawk College

“Most people overestimate what they can achieve in a year and underestimate what they can achieve in ten years” is a quote whose provenance is unknown, though some attribute it to Standord computer scientist Roy Amara. Some people call it the “Gate’s Law.”

The principle is useful for technology market forecasters, as it seems to illustrate other theorems including the S curve of product adoption. The expectation for virtually all technology forecasts is that actual adoption tends to resemble an S curve, with slow adoption at first, then eventually rapid adoption by users and finally market saturation.   

That sigmoid curve describes product life cycles, suggests how business strategy changes depending on where on any single S curve a product happens to be, and has implications for innovation and start-up strategy as well. 

source: Semantic Scholar 

Some say S curves explain overall market development, customer adoption, product usage by individual customers, sales productivity, developer productivity and sometimes investor interest. It often is used to describe adoption rates of new services and technologies, including the notion of non-linear change rates and inflection points in the adoption of consumer products and technologies.

In mathematics, the S curve is a sigmoid function. It is the basis for the Gompertz function which can be used to predict new technology adoption and is related to the Bass Model.

Another key observation is that some products or technologies can take decades to reach mass adoption.

It also can take decades before a successful innovation actually reaches commercialization. The next big thing will have first been talked about roughly 30 years ago, says technologist Greg Satell. IBM coined the term machine learning in 1959, for example, and machine learning is only now in use. 

Many times, reaping the full benefits of a major new technology can take 20 to 30 years. Alexander Fleming discovered penicillin in 1928, it didn’t arrive on the market until 1945, nearly 20 years later.

Electricity did not have a measurable impact on the economy until the early 1920s, 40 years after Edison’s plant, it can be argued.

It wasn’t until the late 1990’s, or about 30 years after 1968, that computers had a measurable effect on the US economy, many would note.

source: Wikipedia

The S curve is related to the product life cycle, as well. 

Another key principle is that successive product S curves are the pattern. A firm or an industry has to begin work on the next generation of products while existing products are still near peak levels. 

source: Strategic Thinker

There are other useful predictions one can make when using S curves. Suppliers in new markets often want to know “when” an innovation will “cross the chasm” and be adopted by the mass market. The S curve helps there as well. 

Innovations reach an adoption inflection point at around 10 percent. For those of you familiar with the notion of “crossing the chasm,” the inflection point happens when “early adopters” drive the market. The chasm is crossed at perhaps 15 percent of persons, according to technology theorist Geoffrey Moore.

source 

For most consumer technology products, the chasm gets crossed at about 10 percent household adoption. Professor Geoffrey Moore does not use a household definition, but focuses on individuals. 

source: Medium

And that is why the saying “most people overestimate what they can achieve in a year and underestimate what they can achieve in ten years” is so relevant for technology products. Linear demand is not the pattern. 

One has to assume some form of exponential or non-linear growth. And we tend to underestimate the gestation time required for some innovations, such as machine learning or artificial intelligence. 

Other processes, such as computing power, bandwidth prices or end user bandwidth consumption, are more linear. But the impact of those linear functions also tends to be non-linear. 

Each deployed use case, capability or function creates a greater surface for additional innovations. Futurist Ray Kurzweil called this the law of accelerating returns. Rates of change are not linear because positive feedback loops exist.

source: Ray Kurzweil  

Each innovation leads to further innovations and the cumulative effect is exponential. 

Think about ecosystems and network effects. Each new applied innovation becomes a new participant in an ecosystem. And as the number of participants grows, so do the possible interconnections between the discrete nodes.  

source: Linked Stars Blog 

Think of that as analogous to the way people can use one particular innovation to create another adjacent innovation. When A exists, then B can be created. When A and B exist, then C and D and E and F are possible, as existing things become the basis for creating yet other new things. 

So we often find that progress is slower than we expect, at first. But later, change seems much faster. And that is because non-linear change is the norm for technology products. 

What Era of Computing Comes Next?

By now, all of us are aware that rapid reductions in computing and storage cost, with rapid increases in capability, can enable applications, use cases and revenue models that were not feasible in the past because computing or storage costs precluded them. 

So ridesharing is possible because people have capable smartphones and mobile internet access fast enough to support that use case. Netflix and other video streaming services are possible because digital infrastructure capabilities have been improved at Moore’s Law rates. 

Applied artificial intelligence is among the capabilities that benefit directly from rapid processing improvements. A study shows that, “before 2010 training compute grew in line with Moore’s law, doubling roughly every 20 months.”

But “since the advent of deep learning in the early 2010s, the scaling of training compute has accelerated, doubling approximately every six months,” say professors Jaime Sevilla, Lennart Heim, Anson Ho, Tamay Besiroglu, Marius Hobbhahn and Pablo Villalobos in a study. 

source: Jaime Sevilla, Lennart Heim, Anson Ho, Tamay Besiroglu, Marius Hobbhahn and Pablo Villalobos

source: Jaime Sevilla, Lennart Heim, Anson Ho, Tamay Besiroglu, Marius Hobbhahn and Pablo Villalobos

“Our findings seem consistent with previous work, though they indicate a more moderate scaling of training compute,” the researchers say. “In particular, we identify an 18-month doubling time between 1952 and 2010, a six-month doubling time between 2010 and 2022, and a new trend of large-scale models between late 2015 and 2022, which started two to three orders of magnitude over the previous trend and displays a 10-month doubling time.”

Moore's Law and rapid increases in computing power, with corresponding reductions in price, matter hugely. It allows entrepreneurs to innovate by asking the question “ what would my business look like if computing or bandwidth no longer were barriers?” 

Does anybody doubt that near-zero pricing remains among the biggest business threats in the connectivity business? And does anybody really doubt that Moore’s Law has led to substitute products for telco voice and messaging while diminishing the cost of transporting bits? 

Has bandwidth not increased, in lead markets, at the headline level, at about the rate Moore’s Law or Nielsen’s Law predicts? 

Edholm’s Law states that internet access bandwidth at the top end increases at about the same rate as Moore’s Law likewise suggests computing power will increase.

Nielsen's Law essentially is the same as Edholm’s Law, predicting an increase in the headline speed of about 50 percent per year. 

The point is that an inflection point has been reached for applied use of artificial intelligence. As we once might have asked “what does my business look like if computing or bandwidth were essentially free,” we now must start asking questions such as “what does my business look like if artificial intelligence is available to use?” 

As when those earlier questions were asked, the cost of training is nowhere near “free.” But neither was computing or bandwidth when the founders of Microsoft and Netflix laid out their plans. 

The most-startling strategic assumption ever made by Bill Gates was his belief that horrendously-expensive computing hardware would eventually be so low cost that he could build his own business on software for ubiquitous devices. .

How startling was the assumption? Consider that, In constant dollar terms, the computing power of an Apple iPad 2, when Microsoft was founded in 1975, would have cost between US$100 million and $10 billion.

Reed Hastings, Netflix founder, apparently made a similar decision. For Bill Gates, the insight that free computing would be a reality meant he should build his business on software used by computers.

Reed Hastings came to the same conclusion as he looked at bandwidth trends in terms both of capacity and prices. At a time when dial-up modems were running at 56 kbps, Hastings extrapolated from Moore's Law to understand where bandwidth would be in the future, not where it was “right now.”

“We took out our spreadsheets and we figured we’d get 14 megabits per second to the home by 2012, which turns out is about what we will get,” says Reed Hastings, Netflix CEO. “If you drag it out to 2021, we will all have a gigabit to the home." So far, internet access speeds have increased at just about those rates.

Everyone has struggled to define what the next era of computing would look like. We might have found our answer, at least relating to nomenclature. Some say we are in the era of cloud computing. Others might prefer mobile computing or web-based computing.

The point is that we left the mainframe, mini-computer, personal computer, client-server eras. Where we are now might be considered the internet, web, cloud-based or mobile era. We have not yet agreed on a specific term. 

What comes next might well be the AI era.

Name One Legacy Firm That Has Actually "Digitally Transformed" its Business Model

Name one legacy entity that really has "digital transformed itself." Note that better return on investment; happier customers; happier employees; better productivity or higher market share, often said to be ways to meassure digital transformation success, actually can be used to measure success of all other efforts to make any business run better.

If any firm cites better performance on those and other metrics because it uses "digital" or "information technology," how is that in any way different from past applications of IT to improve performance?

In other words, is anybody really pursuing digital transformation, or simply spending more on information technology than they used to, to do more things "using the internet" or "online" or "faster."

Many note that “Digital transformation” (DX) efforts fail at about 70 percent rates. In truth, failure to achieve the fullest and deepest meaning of digital transformation might be virtually 100 percent. 

Many argue that DX is, in principle, different from earlier uses of information technology, which were mostly about efficiency and automation. That might be overstating matters, but failure rates for IT projects are often as high as 70 percent.

Perhaps we should simply admit that change efforts fail most of the time, in any sphere. 

Perhaps we also should admit that what people now call “DX” is not what most entities are attempting.

Digital transformation is almost-always said to involve big changes in culture, technology, external and internal processes to achieve a revolution in business models. Indeed, the term “transformation” virtually requires it. 

Most often, most entities saying they are engaged in digital transformation actually are doing something else. Which is to say, they are applying IT mostly to improve or modify existing processes, without fundamentally changing their revenue models, customers or value chains. 

The point is that, no matter what they say, most entities claiming they are doing “digital transformation” really are not doing so. They are applying digital technology, yes. Trying to improve customer experience; product features; response times and efficiency. 

Few really aim to revolutionize their revenue models, change customers, sell products that are not what their legacy entails. 

Digital transformation is said to be an effort to change business models, while applied information technology earlier was mostly about efficiency and automation. 

That claim, like most generalizations, has to be qualified. 

One would be hard pressed to argue that Apple Computer and Microsoft in the early days were engaged in applying technology to improve the efficiency of their businesses. Instead, they were doing transformation: using technology to create new business models and products. 

More recently, one can make the same argument about Amazon, eBay, Airbnb, Uber, Meta, Netflix or Alphabet: each is a transformation, not an effort to be more efficient, in a direct sense. 

To the extent the definition of DX is true, the former aims to create new and different value; new products and services. In this view, earlier uses of applied information technology mostly aimed to produce lower operating costs. 

In a sense, this is the classic distinction between “effectiveness” and “efficiency.” Some might argue that earlier uses of information technology mostly aimed to “do things faster and at less cost.”

In that analogy, digital transformation aims to “do the right things, not existing things faster.”

Still, it is arguably correct to say that most organizations and firms will never move beyond the older “efficiency” focus, because most entities will never really change their business models. They will keep doing what they have been doing, in that regard. 

If one believes that the crucial attributes of DX are changed business models, and not “merely” better customer experience, profit margins, product features and attributes, then few firms will ever achieve DX. 

source: ElevateIQ

So perhaps it is not surprising that 70 percent to 74 percent of DX projects and efforts fail. 

Of the $1.3 trillion that was spent on digital transformation--using digital technologies to create new or modify existing business processes--in 2018, it is estimated that $900 billion went to waste, say Ed Lam, Li & Fung CFO, Kirk Girard is former Director of Planning and Development in Santa Clara County and Vernon Irvin Lumen Technologies president of Government, Education, and Mid & Small Business. 

That should not come as a surprise, as historically, most big information technology projects fail. BCG research suggests that 70 percent of digital transformations fall short of their objectives. 

From 2003 to 2012, only 6.4 percent of federal IT projects with $10 million or more in labor costs were successful, according to a study by Standish, noted by Brookings.

source: BCG 

IT project success rates range between 28 percent and 30 percent, Standish also notes. The World Bank has estimated that large-scale information and communication projects (each worth over U.S. $6 million) fail or partially fail at a rate of 71 percent. 

McKinsey says that big IT projects also often run over budget. Roughly half of all large IT projects—defined as those with initial price tags exceeding $15 million—run over budget. On average, large IT projects run 45 percent over budget and seven percent over time, while delivering 56 percent less value than predicted, McKinsey says. 

Beyond IT, virtually all efforts at organizational change arguably also fail. The rule of thumb is that 70 percent of organizational change programs fail, in part or completely. 

The big observation, however, is that digital transformation--in the sense of new business models--will rarely succeed, in the broad sense of creating entirely-new revenue models. A firm that makes its money selling autos and trucks will rarely become something else, no matter how much technology is embedded in the business. Airlines never become something else, no matter how intensive their IT efforts. 

Name any legacy firm, in any industry, that has truly “transformed” its business model by becoming something else. Try it. You will find scarcely a handful of firms that could make the claim. Perhaps you cannot name even one firm that has achieved a transformation of business models. 

Most firms can only say they are better able to customize or personalize, extend modes of customer interaction or change processes faster. Few, if any, legacy firms can make the claim they now sell to different customers, earn their money in different ways, by selling different types of products.

AI Will Bring New Competitors, Alter Valuations

So far, artificial intelligence has been more a function for most firms than a direct revenue source. Nvidia and AMD are among firms that actually book revenue driven by processing needs for AI. Other firms now earn revenue from supplying AI enablement for software and information or communications products. 

One common outcome of deregulation and technology change in any industry is that non-traditional competitors will enter markets. That is why “new” contestants from “outside” the legacy industry always will appear. 

So Netflix and other video streaming firms emerge as competitors to linear video providers. Amazon emerges to compete with Walmart and Target. 

AI virtually certainly is going to allow new contestants to enter existing markets, just as much as it allows legacy competitors to craft new products, add more value or change their roles in existing ecosystems.

But existing and legacy providers have done so in the past, sometimes because of deregulation; sometimes because of new technology. Cable TV companies entered the voice business, started serving business customers, became internet service providers, then mobile service providers. Some also became content owners and producers. 

Most of those developments were based on deregulation. But technology also played a role, as cable operators were able to use optical fiber and processing technologies to create full communication networks from former "TV delivery" networks.

They also then became conglomerates in terms of their revenue contributors. 

Conglomerates are hard to value, which is one reason financial analysts prefer “pure-play” assets. And more firms in the internet ecosystem seem likely to take on new and different roles in the ecosystem, with more-complicated revenue and valuation profiles. Alphabet, for example, seems generally valued in the same way as content or app firms, though it also earns revenue from products also sold by AT&T, mobile firms, ISPs and data centers.

As Amazon Web Services revenue is valued differently from Amazon e-commerce sales, so a growing range of firms have multi-product strategies that cross traditional industry lines. 

Financial analysts often have coverage of “technology, media and telecom,” or TMT. Larger firms are able to specialize in each of the areas, as each of those segments have distinct business drivers, valuation metrics and roles in the internet ecosystem. 

Consider the component companies in the S&P 500 “Communications” index. It includes firms as disparate as Activision Blizzard and Electronic Arts; Disney and Live Nation Entertainment; Comcast, AT&T and T-Mobile; Alphabet and Meta Platforms; Omnicom and Interpublic.

Nobody would argue they are all in the same segment of the internet business. Fox is a content producer and distributor, as is Take-Two Interactive. Live Nation is a concert producer. 

Omnicom and Interpublic are advertising agencies. Match Group is a dating site while Verizon is a connectivity provider. 

Nobody views all those firms as being in the same part of the ecosystem, or having comparable roles. At the moment, and with the caveat that the ratios shift over time, the price-earnings ratio of the index has relatively recently been about 18. 

source: Sather Research 

Looking only at relative differences in ratios, and not the absolute figures, the single index includes firms and industry segments with different ratios.

Ratios for mobile, ISP or telco firms averaged closer to 14.75. And the ratios within the sub-group vary as much as the overall ratio between segments of the index. Mobile assets might be valued differently from fixed network assets, for example, leading “diversified” firms with both types of assets to have a blended valuation. 

At least relatively recently, AT&T might have a P/E ratio in the 11 range; Verizon possibly in the 14 range while T-Mobile is 16 or so. Comcast might have a 21 ratio while Charter carries a 24 ratio. 

The ratios for non-telco firms were higher, at perhaps 22 or so. The advertising agency firms were individually valued between 19 and 22.

Application or content firms in the index had an average P/E of about 30, though very-high ratios of Netflix and Amazon do skew the average. Disney, Meta and Alphabet arguably represent the more-typical case, with P/E ratios in the 20 to 35 range. 

Valuation always includes many elements, ranging from growth rates to firm size, perceived competition or moats, geography or existence of possible catalysts. More transitory issues, such as recessions, bank panics or new catalysts such as applied artificial intelligence (ChatGPT, generative AI, large language models) also have an impact. 

The larger point is that once firms in the index begin to take on multiple roles, and once those roles produce significant revenue, valuation gets more complicated, since each role (segment) tends to have a different range of values.