TFF #011 - ChatGPT-4 Wrote This Newsletter For You (Mostly)

Read Time - 7 Mins

Hello, and welcome to a slightly different version of the Tech for Finance newsletter.

It’s my wife’s Birthday today, and with the release of ChatGPT-4 this week, I wondered whether it could write my newsletter for me, freeing my time up to be spent with her…

So I asked ChatGPT to write me a newsletter, explaining all different types of AI, their use cases for finance pros, and some example tools.

SPOILER ALERT

I won’t be doing it again, because:

1. It doesn’t sound like me

2. It doesn’t understand you (my audience) and the way you speak

3. Some of the tools it recommended aren’t finance specific, with some requiring an IT Team / Developer or Data Scientist to implement

Here’s what it produced, and if you keep reading, I’ll show the prompts that I used at the end.

—

Tech for Finance Newsletter - March Edition

Hello Financial Controllers and CFOs! Welcome to the March edition of the Tech for Finance newsletter. In this edition, we're diving deep into the world of Artificial Intelligence (AI) and how it can revolutionize your financial operations by automating mundane tasks, so you can focus on strategic decision-making. We have divided this newsletter into three sections: types of AI, use cases for finance professionals, and open-source tools to get you started. Let's dive in!

Section 1: Types of AI

1.1 Rule-based Systems

A rule-based system is like a meticulous accountant who follows strict guidelines. This type of AI makes decisions based on a predetermined set of rules that programmers define, providing a dependable and straightforward approach.

1.2 Machine Learning (ML)

Machine learning is like an observant analyst who learns from historical data. This type of AI uses data and algorithms to teach itself how to make decisions or predictions, becoming more adaptable and improving its performance over time.

1.3 Deep Learning (DL)

Deep learning is like an insightful data scientist who can uncover hidden patterns. This AI subtype uses neural networks—similar to the human brain's structure—to process and analyze vast amounts of data, excelling at tasks like image and speech recognition.

1.4 Natural Language Processing (NLP)

Natural language processing is like an expert linguist who can understand and interpret human languages. NLP allows AI systems to comprehend and generate human-like language, enabling them to read documents, answer questions, or even compose emails.

1.5 Reinforcement Learning (RL)

Reinforcement learning is like a strategic planner who learns from trial and error. In RL, AI systems learn to make decisions based on rewards and penalties, improving their performance over time by adjusting their strategies to maximize rewards.

Section 2: Use Cases for Financial Controllers and CFOs

2.1 Rule-based Systems

Rule-based systems are perfect for financial controllers and CFOs who need to follow strict regulations or specific business rules. They can be used to automate tasks such as:

Compliance checks: Ensure financial transactions and reports adhere to relevant laws and regulations.
Expense approval: Assess and approve expense reports based on predetermined criteria like budget limits and spending categories.

2.2 Machine Learning

Machine learning can help financial controllers and CFOs analyze historical data and discover hidden patterns to make better decisions. Some use cases include:

Forecasting: ML algorithms can predict future revenue, expenses, and cash flow based on historical data.
Anomaly detection: ML can identify unusual financial activities, such as unexpected spikes in expenses or irregular transactions.

2.3 Deep Learning

Deep learning is ideal for financial controllers and CFOs who work with large and complex data sets. Some use cases are:

Document classification: DL can automatically categorize and organize financial documents, such as invoices, receipts, and contracts.
Advanced analytics: DL can uncover complex patterns in financial data, providing deeper insights for decision-making.

2.4 Natural Language Processing

NLP can help financial controllers and CFOs automate tasks that involve processing human language. Examples include:

Automated report generation: NLP can summarize financial data and generate easy-to-read reports, such as income statements and balance sheets.
Intelligent search: NLP-powered search engines can quickly retrieve relevant financial documents and information based on natural language queries.

2.5 Reinforcement Learning

Reinforcement learning can be utilized by financial controllers and CFOs to optimize decision-making and improve efficiency. Some use cases are:

Resource allocation: RL can help optimize the allocation of financial resources, such as budgeting for different departments or projects.
Process optimization: RL can identify bottlenecks and inefficiencies in financial processes, such as month-end close or purchasing, and suggest improvements to streamline operations.

Section 3: AI Tools Built for Finance Teams

3.1 Rule-based Systems

1. Alteryx: A powerful analytics platform that enables finance teams to create rule-based workflows for data processing and automation. Website: https://www.alteryx.com/

2. UiPath: A leading Robotic Process Automation (RPA) platform that helps finance teams automate rule-based tasks, such as data entry and validation. Website: https://www.uipath.com/

3. FICO Blaze Advisor: A rule management system designed to help finance teams automate decision-making processes based on predefined business rules. Website: https://www.fico.com/en/products/fico-blaze-advisor-decision-rules-management-system/

3.2 Machine Learning

1. DataRobot: A machine learning platform that empowers finance teams to build and deploy predictive models for tasks such as forecasting and risk assessment. Website: https://www.datarobot.com/

2. H2O.ai: An open-source machine learning platform that offers finance teams a wide range of ML algorithms for data analysis and predictions. Website: https://www.h2o.ai/

3. RapidMiner: A user-friendly data science platform that enables finance teams to create ML models for tasks such as fraud detection and customer segmentation. Website: https://rapidminer.com/

3.3 Deep Learning

1. IBM Watson Studio: A comprehensive AI platform that allows finance teams to build, train, and deploy deep learning models for tasks such as document classification and advanced analytics. Website: https://www.ibm.com/cloud/watson-studio/

2. KNIME: An open-source data analytics platform that supports deep learning, enabling finance teams to create and deploy custom DL models for their specific needs. Website: https://www.knime.com/

3. Keras: A user-friendly deep learning library that can be integrated with financial applications to enable advanced analytics and data processing. Website: https://keras.io/

3.4 Natural Language Processing

1. Google Cloud Natural Language: A powerful NLP service that can help finance teams automate tasks like report generation, intelligent search, and sentiment analysis. Website: https://cloud.google.com/natural-language

2. Amazon Comprehend: A fully managed NLP service that enables finance teams to extract insights from unstructured text data, such as financial reports and customer feedback. Website: https://aws.amazon.com/comprehend/

3. Lexalytics: A text analytics platform that provides finance teams with NLP capabilities for tasks like document classification, sentiment analysis, and entity extraction. Website: https://www.lexalytics.com/

3.5 Reinforcement Learning

1. OpenAI: An organization that offers research and tools in reinforcement learning, which finance teams can leverage to optimize resource allocation and process improvement. Website: https://www.openai.com/

2. Vowpal Wabbit: An open-source reinforcement learning platform that enables finance teams to create RL models for tasks like process optimization and decision-making. Website: https://vowpalwabbit.org/

3. Ray: A distributed computing platform that provides reinforcement learning libraries, allowing finance teams to build and deploy RL models for optimizing financial processes. Website: https://ray.io/

We hope you found this newsletter insightful and are now better equipped to explore the world of AI in finance. By leveraging these AI technologies and tools, you can automate mundane tasks, optimize decision-making, and focus on the strategic aspects of your job. Until next time, happy automating!

Best regards,

The Tech for Finance Team

—

How did I use ChatGPT-4 to produce this?

This was probably the third or fourth iteration before I gave up.

Here are the prompts:

Prompt 1

Act as a content writer producing a newsletter for their business. Your business is called Tech for Finance, and your value proposition is helping finance pros automate boring stuff so they can focus on exciting stuff.

In 2000 words, produce a newsletter that covers the following topics:

1/ An explanation of different types of AI, listed as separate paragraphs. I should include a wide range of AIs, and include metaphors to make the concepts easy to understand

2/ The use cases for each of these AI technologies for finance professionals

3/ Example tools for each use case as well as the website URL. The tools should be opensource, and without a steep learning curve.

Make sure your response has a dedicated section for each on of the above topics, and is written at an 8th grade reading level.

Output

Produced a load of content that applied to stock market traders which is irrelevant to you guys…

Prompt 2

Re-write this so it is aimed at people like Financial Controllers or CFOs, not stock traders.

Use cases also need to apply to areas like, month-end close, purchasing, document management, reporting, forecasting, analytics etc etc

Output

Better, but as you can see, substituted finance pros, for Financial controllers and CFOs haha! Who writes like that?!

The tools it recommended also weren’t good enough, and too techy.

Prompt 3

Change section 3 to tools that are built for finance teams, and include at least 3 example for each type of AI

Output

What you see above.

TL;DR

1/ I won’t be using GPT-4 to write my whole newsletter again, but it could be useful for research

2/ I’m maybe not as good as writing prompts as I should be. Maybe I should have provided more context

3/ At least you got a very brief overview of the different types of AI and loose benefits to you guys as finance pros

That’s it for this week.

Speak soon.

Adam

—

If AI is a priority for you right now.

I’ve 2 things that might help:

1. If you’re looking for guidance - A free 30 min general chat with me about what tech could help you. You can book a time here.

2. If you’re looking to upskill on ChatGPT - My friend Nicolas Boucher has just released a 92 page guide on ChatGPT with 22 real uses cases for finance pros + step by step instructions. I particularly like the examples where he uses ChatGPT to communicate finance ideas to non-finance professionals. You can get your copy here.

tech for finance