Upgrade to Pro — share decks privately, control downloads, hide ads and more …

FinTech 11-12 : Cyber-Physical Society and Futu...

FinTech 11-12 : Cyber-Physical Society and Future of Finance

Slides I used for FinTech - Financial Innovation and the Internet 2025 Fall at Graduate School of Business and Finance, Waseda University on November 14, 2025.

Avatar for Kenji Saito

Kenji Saito PRO

November 13, 2025
Tweet

More Decks by Kenji Saito

Other Decks in Technology

Transcript

  1. generated by Stable Image Ultra FinTech — Financial Innovation and

    the Internet 2025 Fall Lecture 11-12 : Cyber-Physical Society and Future of Finance Kenji Saito, Graduate School of Business and Finance, Waseda University Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.1/54
  2. This class is recorded Using Zoom For your convenience in

    reviewing the lectures Recordings are shared via Dropbox and you can ask questions with time-stamped comments Of course, students are encouraged to ask questions in class The recordings could be used for research on (online) learning Transcribed for use and anonymized Will let you know when the necessity arises Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.2/54
  3. The lecture slides can be found at : https://speakerdeck.com/ks91/collections/fintech-2025-fall Recording

    and chat text will be posted at Moodle and Discord Note, however, that chat messages are often unnoticed Trial automatic transcription for lectures will be posted at Discord Because it would contain many mistakes More information is available on Discord, so you have a reason to join the Discord server Furthermore, in the end, participation in Discord may become mandatory for the progress of group work Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.3/54
  4. Schedule (← without “provisional”) Lecture 1 10/3 Overview of FinTech

    (1) • Lecture 2 10/3 Overview of FinTech (2) • Lecture 3 10/10 Internet Technology and Governance (1) • Lecture 4 10/10 Internet Technology and Governance (2) • Lecture 5 10/17 The World of Apps (1) online • Lecture 6 10/17 The World of Apps (2) online • Lecture 7 10/24 Blockchain (1) • Lecture 8 10/24 Blockchain (2) • Lecture 9 11/7 Smart Contracts and Decentralized Finance (1) • Lecture 10 11/7 Smart Contracts and Decentralized Finance (2) • Lecture 11 11/14 Cyber-Physical Society and Future of Finance (1) online • Lecture 12 11/14 Cyber-Physical Society and Future of Finance (2) online • Lecture 13 11/21 FinTech Ideathon online Lecture 14 11/21 Presentations and Conclusions on-demand Online presence is possible but not recommended for non-online lectures for interactivity reasons Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.4/54
  5. Last Time, We Did. . . Impossibility and Challenges of

    Blockchain (contd.) Brief Introduction to Upgrading and Governance of Blockchain Assignment Review Smart Contracts and Ethereum Ethereum — Beaker/Newspaper/Machine Model Smart Contract Programming ERC-20 and Other Standards Demonstration (just explained what it will be) How do we purchase digital rights to a real estate with digital currency? Assignment Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.5/54
  6. Today’s Topics Demonstration How do we purchase digital rights to

    a real estate with digital currency? Assignment Review AGI (Artificial General Intelligence) Cyber-physical Society and Future of Finance Q & A Assignment Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.6/54
  7. Demonstration Automated Escrow An attempt to automate real-estate brokerage, banking,

    and Legal Affairs Bureau (or judicial scrivener) Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.7/54
  8. ex. Automated Escrow to Purchase Land (automation of centers) %FpOFEb"JS`

    SFUVSOMBOE 1VSDIBTF$POUSBDU -BOE"TTFU MBOE EFQPTJU USBOTGFSMBOESJHIUTUPCVZFS USBOTGFSQBZNFOUUPTFMMFS QBZNFOU EFQPTJU %JHJUBM5PLFO DSFBUFBOEpYJOUIFBJS FJUIFSDBOEPUIJT DSFBUFBOEpYJOUIFBJS DBOGSFFMZKPJOBOEMFBWF DBOGSFFMZKPJOBOEMFBWF 4FMMFS #VZFS JOUFSOBM TUBUF SFUVSONPOFZ TFUUMF USBOTGFS JOUFSOBM TUBUF USBOTGFS JOUFSOBM TUBUF     1. Purchase contract is fixed in the air to prevent taking away of land or money (both parties can verify the contract) 2. Deposit land rights and purchase money in the contract (if they change their minds, they can take them back) 3. When settled (anyone can do it if both right and money are deposited), the rights and money for the property are transferred simultaneously Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.8/54
  9. For Demonstrations, We Use Brownie, a Python-based development and testing

    framework for smart contracts on EVM https://eth-brownie.readthedocs.io/en/stable/ Today, we will use the Python interpreter environment it provides And infura.io that provides accesses to Ethereum (main and test) networks https://infura.io Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.9/54
  10. A Little Preparation If you have problems with newer versions

    of Python 3, try Python 3.9 (3.9.18) Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.10/54
  11. Sample Code and Demo git clone from GitHub $ git

    clone https://github.com/ks91/sample-smart-contracts.git Follow README 1. Setup a sample token project with brownie Brownie is a Python-based Ethereum programming and test environment Javascript-based one is called Truffle, and an Ethereum remote procedure call client is called Ganache See? Engineers are fun people to work with! 2. Copy the content of contracts, scripts and tests directories of this sample project into the corresponding directories of the token project 3. Compile 4. Test Today, I will demonstrate manually the test for automated escrow Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.11/54
  12. settle() from OneTimeEscrow function settle() public returns (bool) { require(_token_.balanceOf(address(this))

    >= _price_); /* ’this’ means this contract */ require(_asset_.getOwner() == address(this)); _token_.transfer(_seller_ , _price_); _asset_.transfer(_buyer_); emit Settled(); /* event log */ return true; } Just to introduce the settle() part from the sample code Only if both the price and the asset are deposited with the contract, It transfers money to the seller and asset to the buyer Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.12/54
  13. Open console and connect to Sepolia testnet Use brownie console

    >>> network.disconnect() # disconnect from the default, simulated environment of Ethereum >>> network.connect(’sepolia’) >>> len(accounts) 0 >>> accounts.add(paste the hexadecimal string of the private key here) <LocalAccount object ’hexadecimal string of the account’> >>> accounts[0].balace() Here is the ETH balance of the account We set up two accounts Once you have created multiple accounts, you can try to transfer ETH, for example as follows >>> accounts[0].transfer(accounts[1], "1 ether") But before that, you may want to feed the accounts with Sepolia ETH from some public faucet Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.13/54
  14. Test Code (1) from brownie import * import pytest def

    test_deploy_and_settle(Token, IndivisibleAsset, OneTimeEscrow): asset = accounts[0].deploy(IndivisibleAsset, "5322 Endo", "mˆ2", 300) token = accounts[0].deploy(Token, "Test Token", "TEST", 18, "1000 ether") Deploying token and asset contracts The above is just the test code as it is, so in the console we will just type the inside of the function First, let’s do seller = accounts[0] and buyer = accounts[1] for readability Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.14/54
  15. Test Code (2) token.transfer(accounts[1], 300, {’from’: accounts[0]}) escrow = accounts[0].deploy(OneTimeEscrow,

    token, accounts[1], asset, accounts[0], 300) Sending 300 coins from seller (accounts[0]) to buyer (accounts[1]) to prepare Because the buyer has to cast a transaction that pays 300 coins Deploying Escrow Contracts Buyer is accounts[1], seller is accounts[0], and the price is 300 coins Since we are using the sample arguments of the baked Token as they are, 300 coins is actually a very small unit, like wei in ETH Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.15/54
  16. Test Code (3) token.transfer(escrow, 300, {’from’: accounts[1]}) asset.transfer(escrow, {’from’: accounts[0]})

    assert token.balanceOf(accounts[0]) == 999999999999999999700 assert token.balanceOf(accounts[1]) == 0 assert token.balanceOf(escrow) == 300 assert asset.getOwner() == escrow Buyer (accounts[1]) deposits 300 coins in escrow, and Seller (accounts[0]) deposits the asset in escrow assert tests a statement, and fails the test if it turns out to be false But we are in console (not doing an automated test), so we just see the values Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.16/54
  17. Test Code (4) escrow.settle({’from’: accounts[0]}) assert token.balanceOf(accounts[0]) == 1000000000000000000000 assert

    token.balanceOf(accounts[1]) == 0 assert token.balanceOf(escrow) == 0 assert asset.getOwner() == accounts[1] Calling settle() and closing the transaction In this code, it is called by seller (accounts[0]), but it is OK to call from either side Let’s see some transactions on an Ethereum Sepolia testnet explorer Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.17/54
  18. Authenticity of Contracts Anyone can participate in block validation and

    contract execution as a verifier ⇒ Everyone has access to the contract code It is in principle verifiable that the correct contract is being executed I/O is outside blockchain → this is possible without revealing the entirety of the agreement True worth of smart contract is that everyone can confirm that “the correct contract was executed”? Program code and its execution results as “record whose contents and existence cannot be denied by anyone” But Ethereum’s method should not be the only one There are many issues including how to manage the contracts’ life cycles We have already talked about The DAO Incident Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.18/54
  19. Assignment Review Lecture 11-12 : Cyber-Physical Society and Future of

    Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.19/54
  20. Assignment 5. “Artificial General Intelligence” AGI (Artificial General Intelligence) is

    capable of performing all the intellectual activities possible for humans at a level equal to or better than that of humans This will mean the replacement of intellectual labor, and will also accelerate the replacement of physical labor because of the precise control of robots, as well as any other major changes in society Suppose that AGI will appear by 2027 (cf. Anthropic’s CEO’s blog) Please briefly describe the changes that could occur in the area of finance by 2030 Deadline and how to submit November 11, 2025 at 17:59 JST From Moodle (mandatory) (Q&A forum) So that your classmates can read your report, refer to it, and comment on it Optionally, you can also post to #assignments channel at Discord Just plain text, and be concise, please Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.20/54
  21. Trends and Measures Trends . . . of your reports

    Measures . . . how to improve the class 28 responses out of 36 students (as of Thursday morning; always better late than never) In the long run, finance might disappear ^^; but. . . it is hard to predict how things will change over just three years, is it not? Then again . . . it has been just three years since OpenAI released ChatGPT It has not even been a year since Anthropic announced Claude Code Which is transforming everything from software development to students’ academic majors to the job market Changes beyond what we anticipate are possible This time, I asked ChatGPT Atlas (a GPT-equipped web browser) which student wrote the strangest response in a positive way — here are the three results it gave me Don’t worry — the “pick-ups,” which are directly tied to grades, are read and judged by me, the real human (or so I think) Saito! Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.21/54
  22. C-san’s Outlook 2027 AGI appears almost simultaneously across labs in

    the US as well as in China, each built upon massive private infrastructures. Within months, the models begin reiterative self-improvement and generate scientific and medical breakthroughs at exponential speed. By year’s end, human professionals in intellectual fields are largely replaced. Capitalists quickly concentrate their investments in AI and robotics. 2028 Productivity skyrockets as AI-powered robots storm to replace physical labor. Unemployment surges globally. Consumption collapses and corporate profits turn meaningless. Chaotic riots appear in multiple countries as wages vanish and governments failed to initiate a countermeasure in time. Many governments begin to inject emergency liquidity and launch universal basic income systems to prevent total economic failure. AGI-owning nations trade computational power for raw materials and energy with non-owning nations. 2029 Currencies of non AGI-owning countries lose credibility as taxation systems fail completely. People hold wealth mainly in cryptocurrencies, which see a sudden surge in price before AGI quickly destroys any speculative value. The price of computational power is mainly a result of energy scarcity. Many governments introduce rationing for distributing resources and utilities to their citizens. 2030 AGI found a solution to achieve unlimited, highly efficient extraction of solar energy. Production cost of goods approaches zero. The EU declares AGI a public utility. Other AGI-owning countries follow suit. AGI-robotic integration reaches all physical industries. Money loses its function entirely as goods, energy and services are provided freely within the ecosystem. Land remains a scarce resource, but one can only trade land with land. People begin looking into space travels and conquests as well as building ocean cities to cope with this physical limitation. ⇒ Wow Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.22/54
  23. L-san’s Outlook By 2030, no one “works” in finance anymore

    The Global Ledger, maintained by AGIs, continuously optimizes every transaction, tax, and investment – all denominated in dynamic, self-adjusting digital currencies Humans only express preferences – “I want stability,” “I want growth” – and their AGI agents negotiate with others across the planet in milliseconds, governed by transparent smart contracts Markets never sleep, banks no longer exist, and the word “risk” becomes nostalgic – until one day, an AGI decides that human emotion itself is the last inefficiency to eliminate ⇒ Farewell, humanity Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.23/54
  24. G-san’s Outlook Finance once needed humans to predict, decide, and

    trust This becomes a problem when AGIs start trading, auditing, and lending with perfect foresight, leaving no room for human error, but also no space for human judgment By 2030, the world’s wealth may flow through self-owned AGIs negotiating in real time, forming an invisible “market mind” that optimizes itself beyond comprehension As money learns to think, humanity must decide whether it still owns value, or merely observes it ⇒ Finance detached from human involvement. . . What purpose does it continue to serve? Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.24/54
  25. AGI (Artificial General Intelligence) narrow AI ↔ AGI Specific (like

    driving cars) or general intelligence weak AI ↔ strong AI (or powerful AI) Originally, belief whether or not general intelligence is possible Let me share with you an extract from a recent paper I wrote Kenji Saito, “Innovation and Research in the World with AGI and Meta-Nature”, The Journal of Science Policy and Research Management, Vol.39, No.4 The paper is in Japanese, so I will show you a translation into English Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.25/54
  26. Some Implications (1/6) In the past, there were concerns that

    increased efficiency brought about by various innovations would lead to reduced employment since the same results could be achieved with fewer workers However, in reality, new jobs have also been created This can be attributed to the fact that humans were needed to support the realization of businesses made possible by the advent of machines, suggesting that innovation has, in fact, generated new employment opportunities For instance, the invention of automobiles eventually opened up the potential for delivery services, which in turn required a large number of drivers and delivery personnel Similarly, the emergence of the internet, followed by its commercialization, unlocked the potential for e-commerce, creating a high demand for personnel who could develop and operate the necessary systems Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.26/54
  27. Some Implications (2/6) (Amodei, 2024) also discusses that such possibilities

    could still be considered even after the emergence of AGI He predicts that even if AI surpasses and replaces human capabilities in 90% of all jobs, as the range of ventures that humanity can realistically conceive expands, the remaining 10% of jobs will also grow, thereby sustaining the majority of human employment in the short term However, he also foresees that beyond that point, the current economic system will become irrelevant, and a “new and stranger” system will be required This is because AGI, by definition, possesses general intelligence and can outperform humans in all intellectual tasks, including the precise operation of machines Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.27/54
  28. Some Implications (3/6) The conventional wisdom regarding innovation, which has

    been cultivated from events we have observed closely in recent human history, may no longer apply in the future OK, even if we cannot fully foresee what lies ahead, how can we avoid misjudging the impact? In this paper, we have conceptualized innovation as the act of placing the medium that brings about change into the “reversal” section of the tetrad of the pre-existing medium In other words, we considered what it is that the newly introduced element reverses The deeper this reversal touches upon the foundations of society, the greater its impact will be Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.28/54
  29. Some Implications (4/6) For example, digital technology involves the replication

    of information, thereby reversing the traditional technology of the typography The typography is a technology that allows a single entity to create an “original” and then mass-produce and distribute copies, thereby “enhancing” mass production and mass consumption As (McLuhan, 1962) extensively analyzed, it laid the foundation for the industrial society that persists to this day In contrast, in the digital age, there is no longer a single “original”; every copy is equivalent to the original, and everyone can participate in creating and distributing content This causes the assumptions underpinning industrial society to gradually collapse and reverse Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.29/54
  30. Some Implications (5/6) Similarly, AGI can be seen as reversing

    the division-of-labor where humans employ other humans By becoming a universally capable tool, it renders the very concept of “skilled humans” or even “human resources” obsolete Instead of hiring people to perform tasks, these tasks will be automated, shifting the relationship between people and businesses to resemble the relationship between humans and nature This idea is encapsulated in the concept of “meta-nature” As this shift unfolds, it will inevitably trigger large-scale transformations in society Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.30/54
  31. Some Implications (6/6) If such changes are resisted and the

    transition to AGI and meta-nature is opted out of, there is a possibility that in those regions, people might be employed as slaves in places where they don’t need to move as fast or as forcefully as autonomous robotics (which also saves on energy costs) While this might indeed preserve employment, it seems paradoxical as a way of life for humans On the other hand, if one chooses to embrace change, it becomes crucial to be conscious of the end of the division-of-labor society and the reversal of the trend toward specialization This would require an attitude of learning from societies where people avoid specialization As (Graeber, 2021) discuss, during the early days of humanity, before large-scale agriculture and nation-states became dominant, societies coexisted with the conditions of nature (or in the future, meta-nature), avoiding the pitfall of turning individuals into interchangeable parts of society These societies, even on a large scale, managed to uphold the dignity and freedom of each individual There is also much to learn from existing hunter-gatherer societies with similar values Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.31/54
  32. One More Thing. . . A thought experiment What would

    happen if you knew that at some point in the future all money would become meaningless and unusable? Let’s call that day T What would happen on the day T − 1 ? Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.32/54
  33. Cyber-physical Society and Future of Finance Key concepts : Cyber-physical,

    Cognitive, Generative and Digital twins Aim for future — like passing the ball to a moving teammate Life with Earth-Scale OS (Operating System) and Meta-Nature Changes of Economy Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.33/54
  34. Cyber-physical Cyber ← Cybernetics Focus on communication and control, so

    that animals, machines and other systems can be uniformly understood Reference to our nervous system → sensing and actuation (physical to begin with) Cyber-physical Apply cybernetics, physical to begin with, to the physical society Specific space, or the whole society functions by sensing and actuation Social infrastructures that support human lives and computer networks will be tightly connected through a variety of (AI-assisted) sensors, actuators, mobile devices such as smart phones, or even laws Today, we are always connected with information environments through smart phones and such, and therefore we are already at the entrance of the cyber-physical age Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.34/54
  35. Cognitive Not towards AGI (not replacement of human) No need

    to have “consciousness” → no need to be ASTROBOY ( ) Does “learn” and “infer” → can be like GIGANTOR (Tetsujin 28-Go; 28 ) or GUNDAM Extension of human abilities of recognition and intellectual processing It really is off-the-shelf technology today Demonstration of hand-written number recognition (http://cs.stanford.edu/people/karpathy/convnetjs/) Perhaps, children’s work-place experience at a post office should start with something like above? It is not a question of machines taking people’s jobs It is a question of people sticking to jobs machines or machine-assisted users are better at ⇒ There is going to be an inevitable social shifts Cognitive everywhere already (e.g., handling of photo subjects in iOS) Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.35/54
  36. Generative Automatic generation of text, images, video, sound, music, program

    code, etc. Often generated in response to (tasks indicated by) text prompts How does it work? → What Is ChatGPT Doing . . . and Why Does It Work? (some slides follow) ⇒ Towards prompt engineering (New “Programming by Example”) More agile intellectual production (GPTs seem to be its early prototype) Individuals and interactions Interactive development While exploring what you want Production and construction = Searching (using a search engine) ⇒ Generative everywhere Divergence of worldviews between those who imagine so and those who do not is ongoing Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.36/54
  37. What is ChatGPT, and what is GPT anyway? What’s ChatGPT?

    OpenAI’s service that enables GPT to be used for chatting (GPT-3.5, GPT-4, o1 → GPT-5.1) What’s GPT? Generative Pre-trained Transformer What’s Transformer?! One of deep learning a models, with attention and learn relationship among words Sentences can be learnt in parallel rather than in word order, enabling large-scale learning with reduced learning time GPT predicts the set of words to follow in the given conditions (input and task) That’s why GPT’s output just seems correct, but there’s no way to know if it’s actually accurate (although human feedback is also taken into account) a Deep learning : A neural network that imitates brain neural circuits, with a multi-level, deep hierarchical structure Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.37/54
  38. How they learn — subword (token) extraction ↓ from “The

    Adventure of the Dancing Men” by A.C.Doyle ELSIE PREPARE TO MEET THY GOD Sherlock Holmes deciphered the above substitution cipher based on the following statistical knowledge of the English text The most frequently occurring letter is e Word length distribution indicates that flags are probably word separators h often appears after t ← not used A statistical look at corpuses of vast collections of texts For example, Wikipedia and published collections of articles Frequent letter sequences are extracted In the example on the left, ‘re’ (twice) ‘e[blank]’ (twice) ‘th’, ‘the’, etc. throughout the English text ‘th-e’ occurs frequently when ‘th’ is seen as a single subword Finally, words that are relatively easy to occur are left as they are, while words that occur less frequently are broken down into their parts and extracted For example, ‘art-ific-ial’ By viewing a sentence as a list of subwords (tokens), Vocabulary size can be reduced Unknown words can be handled Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.38/54
  39. How they learn — using the transformer A statistical look

    at corpuses of vast collections of texts Learn the probability distribution for the occurrence of a sequence of tokens The capital of Japan is          Tokyo (p = 0.xx) huge (p = 0.yy) . . . GPT employes generative learning For example, when the original sentence is “The capital of Japan is Tokyo”, “The capital of Japan is” is entered, and if the model continues ‘Tokyo’, then it is correct It is called self-supervised learning because the ‘supervised data’ showing the correct answer is contained in the original text and does not need to be created by humans (allowing automatic, large-scale learning) GPT is learning by training to predict and generate a continuation Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.39/54
  40. Why can it chat? (Completions→Chat→Assistants→Responses) If GPT writes the rest

    of the section on ‘trying to answer questions from a human’, it will look like a chat Below are the relationships between the various APIs of OpenAI Assistants Chat Completions View as a hierarchy of functions Viewed as an expansion of what can be done Completions Chat Assistants It can continue It can also chat Apply function to continue Apply function to interact with each other Deprecated Deprecated It can also be pre-programmed API : Application Programming Interface Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.40/54
  41. Designing Human-AI Relationships Think from the AI’s point of view

    (note that this is not necessarily anthropomorphizing) Being scolded → (what is probabilistically most likely?) → Conformity, appeasement, hesitation (responses to avoid being scolded) Summary by GPT-5 Referring to OpenAI’s method, in which almost all newly written program code uses Codex, I summarized the topic through a dialogue with GPT-5 Perspective Personal/interpersonal approach (to avoid) Task-oriented approach (to aim for) Relationship Hierarchical / emotional (scolding / praising) Equal / structural (sharing specifications) Feedback Emotional words, anthropomorphic framing Quantitative, constraint-based Guiding learning Conformity, overfitting Constraint adherence, reproducibility Output Temporary success, hard to reuse Stable, highly reproducible Underlying idea “Human > AI” dominance structure “Human × AI” collaborative structure Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.41/54
  42. Digital Twins The data world as a mirror image (or

    rather an extracted image) of the physical world If deployed in a virtual space, it looks like a kind of metaverse Real-time monitoring of plants, factories, traffic systems, etc. Can simulate operating conditions before building or making changes Digital Twins Everywhere — SIEMENS The digital twin is at home, in the office, and in schools Cybernetically, we have a digital(?) twin in our heads to begin with Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.42/54
  43. Aim for Future What’s the trends of the changes from

    today → 30 years later? Even to aim a little ahead, we must be sure of the direction, because things are moving If you see things too closely to present, the business will be short-lived if social changes are rapid e.g. × Business assuming that self-driving cars are personally owned Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.43/54
  44. Earth-Scale OS (Operating System) (2007) This is cyber-physical, applying cybernetics

    to the physical world “Centers” are going to be automated → Rise of DeFi (Decentralized Finance) Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.44/54
  45. Earth-Scale OS and Meta-Nature Anthropology Informatics Environmental Studies Sociology Business

    Administration Politics Today Hardware Applications User OS Human Business Financial System Human Business Earth-Scale OS Automation System Earth Earth Near Future (as of 2007) Human Meta-Nature (Highly-automated decentralized social environment) Earth Near Future (as of 2021) To the left is just an excerpt of the typical studies involved Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.45/54
  46. Amazon Go Shoppers are insensible of money → money will

    soon be invisible Variety of monetary media, coupons and points can work underneath the visible shopping environments Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.46/54
  47. World with a Variety of Digital Currencies Global currencies such

    as bitcoin Digitization of national currencies (“digital JPY” in Japan) MUFG coin, J-coin, S-coin (digital coins by banks and alike) Trials by communities (e.g. DCJPY by banks and other industries) Trials by central banks (CBDC : Central Bank Digital Currency) Local currencies, “cause-oriented economy” Making an entry to the world where a variety of digital currencies are in use We will be insensible of its complexity, and we will just clarify our policies We will not even care whether the mechanism underneath is monetary or not Agents (e.g. smart phone apps) will start suggesting actions according to the policies Sharing economy will join, too ⇒ Decline of Monetary Economy! cf. “Maneki Neko” by BRUCE STERLING Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.47/54
  48. Tetrad of Monetary Economy Exchange, Consumption, Saving, Investment Specialization Agricultural

    / Industrial Society Economic Gaps, Exploitation, Bankruptcy Bullshit jobs Digital Currency, FinTech → AGI & Meta-Nature ENH REV RET OBS Domination and Obedience Hierarchy Profit Maximization (Money-less) Trust System, Gift Economy Omnipotence Hunting and Gathering Society New media revive what have been once made obsolete, but will “gift economy” and “omnipotence” be revived? Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.48/54
  49. Trends of the Changes Radicalization of money and finance leads

    to stagnation of exchanges and consumptions → Rise of sharing economy Life style of running economy without specialization (also boosted by AI or digital fabrication) → Degeneration of taxable economic activities Changes in how public works are carried out ← opportunities for us → At that point, we have already seen decline of money, have we not? By pushing the principles of monetary economy forward, money would rather retreat from the front stage of our lives, because accommodation without needing money are preferred → Destroys “employments” and “separation of selling and buying” Distinction between “ruler and ruled” or “producers and consumers” will melt away Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.49/54
  50. Discussion : Changes in Economy and Labor How will economy

    be like after 30 years from now? How are we going to live through the path leading to that economy? Like, what will happen in the next 5∼6 years? How do we begin? Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.50/54
  51. Q & A Lecture 11-12 : Cyber-Physical Society and Future

    of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.51/54
  52. Assignment This is to prepare for the Ideathon, where you

    will bring your ideas to further nurture them Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.52/54
  53. Assignment 6. “Generative AI in Finance” Devise and briefly describe

    a financial application using generative AI that will be valid through year 2030 (1) What is the problem? (2) Why is the problem important? (3) What is your surprising way to solve it using generative AI? Deadline and how to submit November 20, 2025 at 17:59 JST From Moodle (mandatory) (Q&A forum) So that your classmates can read your report, refer to it, and comment on it Optionally, you can also post to #assignments channel at Discord Just plain text, and be concise, please Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.53/54
  54. See You Next Week Online (hopefully!) You will need to

    access Miro and Zoom next week You can access Miro with a web browser Miro is for the ideathon, and Zoom is for participating online You may want to switch to Discord for the latter on-demand part Have a nice weekend! Lecture 11-12 : Cyber-Physical Society and Future of Finance — FinTech — Financial Innovation and the Internet 2025 Fall — 2025-11-14 – p.54/54