New AI and Computer Systems

A2

New AI and Computer Systems

Introduction

New AI can plan and do work alone. This changes how banks work and how computers are built.

Main Body

Banks want to use this AI. But the AI needs good and clean information. Banks must follow strict rules. They need to know why the AI makes a choice. Many banks are still learning. A study says 57% of banks are preparing for this AI. They are organizing their data now. Computers are also changing. AI now needs more CPUs and more memory. It needs fewer GPUs than before. Meta and AMD are already using this new way. Some people use many small AI models together. These small models work as well as one big model. This is cheaper and faster.

Conclusion

AI needs clean data for banks. It also needs different computer parts to work well.

Learning

🟢 Comparing Things (More & Fewer)

In the text, we see how computer parts change. To reach A2, you need to describe how much of something you have.

1. Use 'More' for addition

  • AI needs more CPUs.
  • AI needs more memory.
  • (Meaning: + CPUs / + Memory)

2. Use 'Fewer' for smaller numbers

  • It needs fewer GPUs.
  • (Meaning: - GPUs)

Quick Guide: More\uparrow (Increase) Fewer\downarrow (Decrease)


🔵 The 'How' of Action

Look at these two phrases from the story:

  • ...do work alone.
  • ...work as well as one big model.

These words tell us how the AI works.

Example:

  • The AI works alone \rightarrow No help from humans.
  • Small models work well \rightarrow They are good.

Vocabulary Learning

plan (v.)
to think about and arrange something to happen
Example:She plans to travel next month.
alone (adj.)
by oneself, without others
Example:He went to the park alone.
changes (v.)
to make something different
Example:The new law changes many people's lives.
banks (n.)
financial institutions that hold money
Example:She works at a bank.
clean (adj.)
free from dirt or impurities
Example:Please keep the room clean.
information (n.)
facts or knowledge about something
Example:The book provides useful information.
strict (adj.)
very serious or exact in rules
Example:The teacher has strict rules.
rules (n.)
guidelines that must be followed
Example:Follow the rules of the game.
choice (n.)
a decision between options
Example:She made a difficult choice.
learning (n.)
the process of gaining knowledge
Example:Learning new skills is important.
data (n.)
facts or statistics collected for analysis
Example:The data shows a clear trend.
memory (n.)
the ability to remember or the amount of information a computer can hold
Example:The computer has a lot of memory.
B2

The Integration of Agentic AI and System Orchestration in Finance and Computing

Introduction

The rise of agentic AI is causing a major change in how financial services operate and how data center hardware is designed.

Main Body

In the financial sector, agentic AI refers to systems that can plan and execute tasks on their own. However, these systems depend on high-quality, well-managed data. Steve Mayzak from Elastic emphasized that the success of these AI tools is limited by the quality of the data they use. Because financial laws are very strict, companies need clear and trackable results to ensure accountability. Furthermore, moving from simple data to complex natural language requires better indexing to prevent information from being lost in different departments. Consequently, adoption is happening slowly; a Forrester study shows that 57% of financial firms are still building the necessary internal skills for full use. At the same time, a new approach to computing called 'orchestration' is changing hardware needs. This method distributes workloads across various channels, which increases the demand for Central Processing Units (CPUs) and memory rather than just Graphics Processing Units (GPUs). Morgan Stanley analysts asserted that agentic AI needs a higher CPU-to-GPU ratio to handle complex functions. This trend is visible in Meta's use of Amazon Graviton CPUs and its partnership with AMD. Additionally, researchers from Vidoc Security Lab and Aisle proved that coordinating several smaller, public models can produce the same results as one advanced model, such as Anthropic's Mythos, in the field of cybersecurity.

Conclusion

The growth of agentic AI currently depends on two main factors: strict data management in finance and a more diverse approach to hardware in computing.

Learning

🧩 The 'Logic Chain' Shift: From Simple to Sophisticated

To move from A2 to B2, you must stop using only simple connectors like and, but, and because. You need to show cause and effect using professional transitions.

Look at how the text builds an argument:

"Because financial laws are very strict... Consequently, adoption is happening slowly."

🚀 The B2 Upgrade: Cause & Effect

Instead of just saying "so," try these structures found in the text:

  • Consequently \rightarrow Use this to show a direct result. (A2: "It rained, so I stayed home." \rightarrow B2: "The weather was terrible; consequently, I decided to stay home.")
  • Furthermore \rightarrow Use this to add a second, stronger point. (A2: "Also, the car is fast." \rightarrow B2: "The car is incredibly fuel-efficient; furthermore, it is the fastest in its class.")
  • Rather than \rightarrow Use this to show a preference or a shift in direction. (A2: "I want water, not juice." \rightarrow B2: "The company is investing in CPUs rather than just GPUs.")

🛠️ Practical Application: The 'Result' Pattern

In the article, the author uses a specific flow: Constraint \rightarrow Requirement \rightarrow Result.

  1. Constraint: Strict laws.
  2. Requirement: Trackable results.
  3. Result: Slow adoption.

Your Goal: When speaking or writing, don't just list facts. Connect them. Instead of saying "I studied hard. I passed the exam," say: "I studied consistently for three months; consequently, I passed the exam with ease."

Vocabulary Learning

integration
The act of combining different parts into a single system.
Example:The integration of the new app improved user experience.
orchestration
The arrangement and coordination of multiple elements to work together.
Example:The orchestration of the project required clear communication.
agentic
Capable of acting independently and making decisions.
Example:Agentic AI can plan tasks without human help.
accountability
The responsibility to explain actions and decisions.
Example:The company’s accountability was questioned after the data breach.
indexing
The process of organizing data so it can be quickly retrieved.
Example:Proper indexing speeds up database queries.
adoption
The act of beginning to use something.
Example:The adoption of electric vehicles is increasing.
approach
A particular way of doing something.
Example:His approach to learning is hands‑on.
method
A systematic way of doing something.
Example:The method used in the experiment was rigorous.
distributes
To spread or allocate across different places or people.
Example:The software distributes tasks among multiple servers.
workloads
The amount of work assigned to a system or person.
Example:High workloads can lead to burnout.
demand
The need or desire for something.
Example:There is high demand for cloud services.
ratio
A quantitative comparison between two values.
Example:The CPU‑to‑GPU ratio affects performance.
complex
Having many interconnected parts or difficult to understand.
Example:Complex systems require careful monitoring.
functions
Tasks or operations performed by a system.
Example:The software performs several functions.
partnership
A cooperative relationship between entities.
Example:Their partnership increased market reach.
proven
Demonstrated to be reliable or effective.
Example:Proven techniques reduce errors.
coordinating
Arranging different elements so they work together.
Example:She is coordinating the team’s efforts.
models
Representations or simulations used for analysis.
Example:Machine‑learning models predict outcomes.
advanced
Highly developed or sophisticated.
Example:Advanced technology speeds up processing.
cybersecurity
Protection of computer systems from attacks.
Example:Cybersecurity threats are growing worldwide.
C2

The Integration of Agentic AI and Orchestration Architectures within Financial and Computational Infrastructures

Introduction

The emergence of agentic AI is driving a systemic shift in both the operational requirements of financial services and the hardware architectures of data centers.

Main Body

Within the financial services sector, the deployment of agentic AI—defined as systems capable of autonomous planning and execution—is predicated upon the establishment of authoritative, governed data stores. Steve Mayzak of Elastic asserts that the efficacy of these systems is constrained by the quality and availability of underlying data. Given the stringent regulatory environment, there is a critical requirement for deterministic outputs and auditable logic to ensure accountability. The transition from structured data to the processing of complex, unstructured natural language necessitates sophisticated indexing to prevent the fragmentation of information across organizational silos. Consequently, the adoption of these technologies is incremental; a Forrester study indicates that 57% of financial organizations are currently developing the internal capabilities requisite for full implementation. Parallel to these operational shifts, a transition in computational architecture, termed 'orchestration,' is altering the demand for hardware. This paradigm involves the distribution of workloads across multiple processing channels, thereby increasing the relative requirement for Central Processing Units (CPUs) and memory systems compared to the previous reliance on Graphics Processing Units (GPUs). Morgan Stanley analysts suggest that agentic AI will necessitate a higher CPU-to-GPU ratio to manage increased system complexity and tool-use functions. This shift is evidenced by Meta's utilization of Amazon Graviton CPUs and its strategic agreement with AMD. Furthermore, the viability of orchestration is demonstrated in the cybersecurity domain, where researchers from Vidoc Security Lab and Aisle have successfully replicated the results of advanced models, such as Anthropic's Mythos, by coordinating smaller, less advanced public models through standardized workflows.

Conclusion

The trajectory of agentic AI is currently defined by a dual requirement for rigorous data governance in the financial sector and a diversified hardware approach in computational infrastructure.

Learning

The Architecture of Precision: Nominalization and the 'Static' Dynamic

To bridge the gap from B2 to C2, a student must move beyond describing actions and start architecting concepts. The provided text is a masterclass in high-density nominalization—the process of turning verbs and adjectives into nouns to create a formal, objective, and authoritative tone.

🔍 The Linguistic Pivot

Observe the shift from a B2-style narrative to the C2-style academic prose found in the text:

  • B2 approach: "Financial services are changing because agentic AI is emerging, which changes how they operate." (Focus on action and process).
  • C2 approach: "The emergence of agentic AI is driving a systemic shift in both the operational requirements..." (Focus on entities and states).

🛠️ Deconstructing the 'C2 Weight'

In the sentence "The transition from structured data... necessitates sophisticated indexing to prevent the fragmentation of information," we see three heavy-lifting nominals:

  1. Transition (from transitioning)
  2. Indexing (from indexing/to index)
  3. Fragmentation (from fragmenting)

By using nouns, the author removes the need for a human subject (e.g., "Companies are transitioning"), which removes subjectivity and replaces it with institutional authority. The logic becomes an objective truth rather than a corporate observation.

⚡ The 'C2 Upgrade' Matrix

To achieve this level of sophistication, replace causal verbs with noun-phrase drivers:

B2/C1 PhrasingC2 Nominalized EquivalentEffect
Because they are regulated strictly...Given the stringent regulatory environment...Converts a cause into a context.
They need to be able to audit the logic...A critical requirement for auditable logic...Converts a need into a prerequisite.
They are slowly adopting these tools...The adoption of these technologies is incremental...Converts a trend into a measurable phenomenon.

The Golden Rule for C2 Mastery: When you want to sound more authoritative, stop describing who is doing what and start describing which phenomenon is necessitating which requirement.

Vocabulary Learning

agentic (adj.)
pertaining to or exercising agency; capable of independent action
Example:The agentic AI system made autonomous decisions without human intervention.
orchestration (noun)
the coordinated arrangement of multiple components or processes to function together seamlessly
Example:The orchestration of compute resources ensured efficient workload distribution.
deterministic (adj.)
producing the same output from the same inputs each time; predictable
Example:The algorithm's deterministic nature guarantees reproducibility.
auditable (adj.)
capable of being examined and verified, especially for compliance
Example:The system's auditable logs allowed regulators to trace all transactions.
fragmentation (noun)
the breaking up of something into smaller, often disconnected parts
Example:Data fragmentation across silos hampers unified analytics.
incremental (adj.)
occurring in small, successive stages rather than all at once
Example:The deployment strategy was incremental, rolling out features gradually.
paradigm (noun)
a typical example or pattern; a model of thinking
Example:The new paradigm shifted focus from GPUs to CPUs.
necessitate (verb)
to require as a result or condition
Example:The complexity of the system necessitates higher CPU-to-GPU ratios.
viability (noun)
the ability to function successfully or survive
Example:The viability of orchestration depends on robust networking.
trajectory (noun)
the path or course of something over time
Example:The trajectory of agentic AI is moving toward greater autonomy.
diversified (adj.)
varied, incorporating multiple different elements
Example:A diversified hardware approach includes CPUs, GPUs, and specialized accelerators.
rigorous (adj.)
extremely thorough, exact, and careful
Example:Rigorous data governance ensures compliance with regulations.
autonomous (adj.)
acting independently without external control
Example:Autonomous systems can adapt to changing environments.
strategic (adj.)
relating to the identification of long-term goals and the means to achieve them
Example:The strategic partnership aimed to accelerate cloud adoption.
coordinating (verb)
organizing or arranging parts to work together
Example:The team was coordinating the deployment of resources.