As someone who uses language models including Gemini, Perplexity and Claude for various research tasks including some that seek to summarize market trends, I have often found that I get answers that use different sources, which is not unexpected.
What has been unexpected is the frequent refusal of Gemini to provide estimates the other engines do supply. That leads me to believe Gemini, in particular, uses algorithms intended to limit its use in many ways that might expose Alphabet to regulatory or other exposure.
My guess is that Google’s higher regulatory and antitrust exposure, compared to firms supporting Perplexity or Claude, for example, causes the use of guardrails that instruct the chatbot to avoid anything that might be construed as “financial advice,” even when no personally-identifiable information is involved, and the questions relate to industry market sizes, revenues and so forth.
The issue here is not that different training data has been used, that data recency varies or that models use different underlying architectures, algorithms, and fine-tuning techniques. So even when working from similar base data, different models can generate slightly-different conclusions.
Refusals to forecast (like you sometimes see from Gemini) are” typically due to built-in safety protocols designed to prevent the AI from giving unlicensed financial advice, acknowledge the inherent uncertainty of markets, and avoid potential liability and the spread of misinformation,” Gemini itself says, when asked about “refusal to answer” responses.
World is a mobile app including cryptocurrency that is designed to function as a Super App, a multipurpose app (similar to WeChat) combining a social network with commerce and currency support.
People might see significance in Sam Altman's (OpenAI) involvement. Elon Musk likewise is said to be interested in developing a Super App built around X (now merged with xAI).
Such apps typically start with a core service, such as messaging or ride-hailing, and expand to include various other features like payments, e-commerce, social networking,
Originally successful in Southeast Asia and China, there obviously is some interest in whether the same popularity will occur in other markets, especially where banking, payment and e-commerce ecosystems already are robust.
Super Apps also might have succeeded first in markets without a developed application ecosystem. The U.S. app market, for example, is mature and highly competitive. Dominant players already exist in virtually every key vertical that might be part of a single Super App:
* Social Media/Messaging: Meta (Facebook, Instagram, WhatsApp), TikTok, X (formerly Twitter), Snapchat.
* Payments: PayPal, Venmo (owned by PayPal), Cash App (Block/Square), Zelle, Apple Pay, Google Pay, credit card companies.
* Transportation: Uber, Lyft.
* ood Delivery: DoorDash, Uber Eats, Grubhub.
Observers might also note that success so far has been most clear in culturally-homogenous countries, though it remains unclear whether this is a functionally-important issue or not.
Convenience often is said to be the driver behind consumer acceptance of Super Apps and some might note that adoption to date has been strongest in mobile-first countries in Asia where banking and payment systems often are less developed.
As has been the case for digital payment systems generally, countries and markets with well-developed payment and banking systems have not shown the same level of interest. It is likely no accident that Super Apps are not developed or used in North America, Europe and other regions with good banking systems and ubiquitous fixed network internet access availability.
Though "convenience" is said to be the driver of usage, perhaps it is something else that makes Super App usage attractive. A mobile-frist app might make lots of sense for users who are out and about quite a lot, living in highly-urban areas, always with a smartphone and culturally attuned to mobile app behavior rather than also having easy and convenient access to other device form factors (such as personal computers).
How much value does conducting any number of operations or experiences (using social media, consuming content, conducting transactions) within a single app actually provide, in markets where the "best of breed" or preferred providers are separate apps?
It often has been claimed that consumers prefer bundled services such as internet access, video entertainment and communications (mobile and fixed) from a single provider because of "convenience." I have often thought that was incorrect.
It is not the "convenience" of having a single provider of those services, on one bill, as it is the price discounts such bundling provides. "Lower cost" is the value, not convenience.
Similarly, it might be that some value other than mere convenience is the driver of Super App usage. Perhaps where it is successful, the various Super App component capabilities actually are the "best of breed" and preferred experiences.
That said, if the components are "best of breed," or close to it, then the integration of functions (chat, talk, share, schedule, purchase) in a single app might be useful.
Beyond consumer preferences for "best of breed" or "integrated" approaches, there are many potential softer issues that could limit Super App creation in some markets. U.S. antitrust and privacy adfocates would likely oppose any dominant supplier in any single market niche from gaining too much share in adjacent and complementary markets.
Researchers looking at Antropic’s Claude language model 3.5 Haiku find evidence that Claude sometimes thinks in a conceptual space that is shared between languages, suggesting it has a kind of universal “language of thought.”
Perhaps more notably, researchers say Claude “will plan what it will say many words ahead, and write to get to that destination,” which is an extension of the basic language model function of predicting what the next word in a sentence might be.
“Language models are trained to predict the next word, one word at a time,” researchers note. “Given this, one might think the model would rely on pure improvisation.”
“However, we find compelling evidence for a planning mechanism,” they say. “First, the model uses the semantic and rhyming constraints of the poem to determine candidate targets for the next line. Next, the model works backward from its target word to write a sentence that naturally ends in that word.”
Specifically, the model often activates features corresponding to candidate end-of-next-line words prior to writing the line, and makes use of these features to decide how to compose the line.
“We show this in the realm of poetry, where it thinks of possible rhyming words in advance and writes the next line to get there,” researchers note. “This is powerful evidence that even though models are trained to output one word at a time, they may think on much longer horizons to do so.”
All that matters as we try to document and lay bare the “thought” processes used by Claude and other language models. A language model might be a “black box,” but we still need to understand what happens inside the box.
All the angst we sometimes hear notwithstanding, new technology including artificial intelligence always changes learning, teaching and research. Once upon a time the invention of printing enabled mass availability of books. Television enabled distance learning. Personal computers individualized instruction.
The internet broke geographical barriers and made instant knowledge access possible. Smartphones mean we can learn any place, any time. AI might simply enable more personalization and customization of learning or research than has been possible, until now, in the same way that the internet has allowed personalization and customization of content for individuals.
Technology
Era
Impact on Learning
Impact on Education
Impact on Research
Printing Press
15th Century
Enabled mass production of books, increasing literacy
Standardized curricula and made knowledge more widely accessible
Allowed for broader dissemination of research findings
Television & Radio
20th Century
Brought educational content to remote locations
Allowed for distance learning and public education programs
Provided a medium for academic discussions and public dissemination
Personal Computers
Late 20th Century
Allowed for interactive learning via software
Digitalized education and enabled early e-learning
Increased computational power for data analysis
Internet & Search Engines
1990s-Present
Gave instant access to global knowledge
Enabled online courses, MOOCs, and remote education
Made literature reviews and information retrieval faster
Smartphones & Mobile Learning
2000s-Present
Allowed learning anywhere, anytime
Mobile apps and microlearning increased accessibility
Enabled field research and instant data collection
Artificial Intelligence (AI)
2020s-Present
Personalized learning experiences and AI tutoring
Automated grading, chatbots for student support, and data-driven education strategies
AI accelerates data analysis, hypothesis generation, and automates research tasks
Virtual & Augmented Reality (VR/AR)
Emerging
Immersive, experiential learning (e.g., medical simulations)
Enhances classroom engagement with interactive lessons
Enables virtual labs and remote field studies
Blockchain
Emerging
Secure academic credentials and transcripts
Prevents diploma fraud and enhances credential verification
It perhaps always is difficult to value copper access lines when considering an acquisition with the intention of upgrading those lines to fiber access. It might also be somewhat difficult to value fiber lines in neighborhoods and parts of cities, even when there is no intention to buy copper lines and upgrade them.
Without question, though, the “upgrade” analysis is more difficult. For starters, not all lines really are candidates for upgrading. In some cases, most lines might not be candidates. In such instances, the “upgrade to fiber” business plan will hinge on a minority of lines.
Assume that perhaps 35 percent to 45 percent of Lumen Technologies' consumer access lines could be profitably upgraded to fiber.
But assume the hypothetical $5.5 billion purchase price of the Lumen “consumer fiber business” by a buyer such as AT&T is reasonably accurate, and only includes the already-built fiber assets and customers.
Without further details, we are left to wonder what assets are included, but It might be reasonable to conclude that it is a “cherry picked” set of assets not including central offices, voice infrastructure and copper lines.
That might be because the clearest economics are already captured by the existing fiber facilities. Back in 2022 Lumen’s fiber-to-home footprint reached about 27 percent of total access lines. By some estimates it is possible that Lumen or another owner could upgrade between 35 percent to 45 percent of consumer access lines to fiber on a profitable basis.
But by some estimates Lumen might have built most of the lines it can in markets where it would be the first fiber provider. In many cases the business case for upgrading and becoming the second fiber provider in a neighborhood might not be attractive.
In markets where a single provider uses fiber, consumer buy rates can hit 40 percent of locations passed. In a market where Lumen is the second provider, it might only get 20 percent take rates.
The flip side is that more than half of all Lumen’s existing copper facilities likely cannot be upgraded for economic reasons.
And the copper-based business continues to decline. In early 2024, Lumen had perhaps 4.2 to 4.6 million consumer access lines generating revenue. By early 2025, this number is likely to have further decreased to 3.6 to four million consumer access lines used by paying customers.
Access Line Type
Total Lines
Total Consumer Accounts
Total Consumer Access Lines
8,200,000
N/A
Fiber Lines
3,600,000
2,100,000
Copper Lines
4,600,000
1,900,000
Basically, a buyer intending to upgrade Lumen consumer lines is basing that decision on perhaps 2.9 million to 3.6 million out of 8.2 million lines, conservatively. By some estimates, Lumen might already have upgraded as many as 3.6 million lines, though that figure also includes small business lines that are routinely counted in the “mass markets” bucket.
Perhaps there is some revenue to be generated from the copper lines, but it is a declining resource.
Based on a $5.5 billion purchase price, that implies a per-line investment of between $1897 and $1528 for existing fiber lines, possibly not including any copper lines that are theoretically upgradeable.
We must assume that there are two different types of potential buyers. In one camp are firms that see the potential to increase equity value by upgrading copper access facilities to fiber. In another camp are firms that primarily want the incremental revenue. The former includes firms that see eventual asset sales. The latter mostly includes operating firms in the business for the long haul.
If we assume that Lumen would prefer to get out of the consumer mass markets business altogether, a key issue is whether the rest of the consumer business and facilities (central offices, voice infrastructure, non-upgradable lines) would be retained, spun off to another third party if possible, or bundled on a low-cost basis to a potential buyer that really just wants the fiber assets.
It’s messy. For starters, Lumen (or any new owner acquiring the whole mass markets business) probably would continue to be viewed by regulators as a “carrier of last resort,” meaning it would have to keep offering voice services broadly and might also not be allowed to decommission the copper access network.
An owner might argue it could use other technologies (mobile network, for example) to supply voice and lower-speed internet access service, even if it decommissioned the whole copper network. But regulators have resisted such pleas in the past.
The point is that an acquisition of the Lumen mass markets business would be messy. The value is the fiber lines and potential boost in fiber customers. But getting those lines might also entail getting lots of copper lines that actually cannot be upgraded and have declining value.
And if a potential acquirer only wanted the fiber for internet access and other “data” purposes, the central offices and voice infrastructure would not be very helpful. Beyond that, Lumen’s consumer fiber access lines are scattered about in some neighborhoods in many cities. There are no cities with ubiquitous fiber in place.
Of course, it always is possible that a potential acquirer really only wants the fiber-to-home facilities that already are in place (neighborhoods), with no intent to buy copper lines and upgrade them. That’s arguably an easier business case to make, as there is not requirement for additional capital for the upgrades from copper.
Ambient scribes convert verbal patient-provider interactions into structured notes for clinical documentation and, eventually, medical billing. Useful, of course. The innovation saves time. But at least one study suggests the financial impact is unclear.
That is likely to be a recurring issue for many types of artificial intelligence features and apps.
The issue is that faster task completion, reduced human error, and streamlined workflows do not always translate that into immediate financial gains. In fact, financial impact might be neutral to negative at first precisely because time and money has to be spent to implement the solutions.
This perhaps is not unusual for new technology solutions. Enterprise resource planning (ERP) systems also promised efficiencies, such as the ability to generate reports faster. Still, the financial payoff wasn’t instant. As always, firms had to redesign their business processes to scale up.
Likewise, cloud computing cut information technology overhead and boosted agility, but early adoption did not always lead to an immediate financial outcome.
Perhaps AI operational wins are the low-hanging fruit. Customer service chatbots reduce call center workload, but revenue metrics do not automatically improve.
A 2023 Gartner report suggested 60 percent of AI projects improve process metrics, but only 30 percent show clear financial uplift within a year, for example.
Study/Source
Operational Improvement
Potential Revenue/Profit Impact
Gartner Research
IT leaders in mature AI organizations identify business metrics early and use clear attribution strategies3
Not immediately quantifiable
Cleveland Clinic
AI used to predict patient influx and optimize staffing3
