Step 1- My outlook on AI in Finance

In my field, I currently see AI being used mainly to guide people toward investment decisions, especially through models that claim to predict how certain stocks will move. But personally, I feel that this opens the door to huge risks. These AI models depend on the data fed into them, and I believe brokerage firms or large investors could easily influence them, whether intentionally or through biased inputs. People might rely on those outputs, invest in the wrong stocks, push prices up artificially, and then the person behind the manipulation could exit before the stock crashes from real news. So while AI seems helpful, it can unintentionally support speculation and even manipulation if we are not careful.

Looking ahead, especially after interviewing Dr. Nourhan, I’m convinced that AI will soon be able to build full financial models, linking sheets, calculating multiples, building DCFs, simply by providing the financial statements and a few assumptions. If you want a multiple model, AI might just need one competitor’s data to produce a full valuation. This reminds me of my experience at Samcrete: we used to spend days estimating construction costs, researching prices, projecting revenue, and then building a full financial model manually. With the way AI is evolving, I can imagine doing the same thing in minutes by feeding the research and assumptions into a powerful model that generates an entire valuation automatically.

I also see AI currently being used in risk management, especially in detecting unusual transaction patterns or early signs of fraud. Banks and financial institutions rely on machine-learning systems to spot anomalies much faster than humans could. This makes the financial system safer in many ways, but it also raises questions about over-automation and how much decision-making we should entrust to algorithms. Overall, AI in finance feels both exciting and dangerous, and I think the future will depend on how responsibly we choose to use it.

Step 2 Ai Perspective on AI in Finance

The AI tool suggested that AI will be used in finance in highly practical and operational ways, especially in tasks that depend on recognizing patterns, processing large datasets, and creating automated financial models. According to the tool, the most immediate uses include:

1) Automated financial modeling and valuation
AI will automatically build linked financial models such as discounted cash flow models and comparable multiples models using raw financial statements. It will also clean data, detect inconsistencies, and generate sensitivity tables. Analysts will spend more time interpreting results rather than building spreadsheets manually.

2) Real-time investment research
AI will scan earnings calls, filings, market data, and global news to generate automated research briefs. It will detect sentiment, highlight risks, and identify catalysts faster than traditional research workflows.

3) Predictive analytics for market trends
AI will use machine learning to detect early signals of momentum, sector rotation, liquidity stress, currency risk, and corporate distress. These tools will not “predict the future,” but they will enhance probability-based scenario modeling.

4) Fraud detection & risk management
Banks will deploy AI systems that monitor real-time transactions, credit behavior, and abnormal portfolio movements. AI will flag outliers faster than manual risk officers and help prevent internal and external fraud.

5) Personalized financial advice
AI-powered budgeting, investment planning, and savings tools will give individuals tailored recommendations based on spending patterns, income cycles, subscription habits, and credit data.

6) Compliance automation
Regulations like KYC/AML will be automated through AI that can detect suspicious identity patterns, verify documents, and monitor transactions—reducing compliance costs significantly.

7) Banking operations automation
Loan approvals, customer support, document processing, treasury operations, and reconciliation will become more automated, making banks more efficient and lowering operational risk.

prompt : Use one of the AI tools to see what it will suggest are ways AI will be used in finance. Cite the AI with the link and your prompt (1 point)

https://chatgpt.com/share/692f28cd-fdc8-800c-b3a0-aa24c73b8cd3

Step 3 First Interview (Dr Moataz El-Helaly)

I interviewed Professor Moataz El-Helaly, an Accounting professor who teaches data analytics and engages deeply with AI tools in academic and professional finance contexts. I chose him because he understands AI beyond the surface level, works in a field closely connected to Finance, and has hands-on experience with analytics, auditing, and financial reporting.

Professor El-Helaly began by emphasizing that AI is still not properly utilized in academia. He sees two extreme groups among faculty:
• Advocates and enthusiasts, and
• Highly anxious professors who fear AI and avoid engaging with it.

This anxiety creates a “sensitivity barrier”, preventing faculty from learning about AI or exploring its teaching and research potential. Meanwhile, the opposite problem exists for students: they adopt AI too aggressively, mainly as a time-saving shortcut. 

The challenge becomes cyclical: students don’t know how to use AI properly, and faculty members don’t always know enough to guide them. This combination produces resistance from one side and shallow overuse from the other. Despite this, he believes AI has “huge potential,” but only if universities create proper institutional regulation, ethical frameworks, and AI literacy pathways. When I asked him what guidelines he thought professors should follow, he didn’t give a concrete answer. But from the way she spoke, I got the sense that he wanted guidelines that encourage informed AI use rather than restrict it. Maybe he imagined something like educational sessions or even a dedicated course, similar to the critical thinking classes students already take, that would train both students and professors on how to use AI responsibly and ethically. (This part of the conversation happened outside the recorded interview.) Tools will change fast, he said, so the priority should be building critical thinking, evaluation skills, and conceptual understanding, not memorizing specific platforms.

In his own teaching, especially data analytics, he uses AI-generated synthetic data where AI is basically extracting data for him but he emphasized that this is the only thing that he can use it for and does not require more tasks due to possibility of delusions. But he stressed that this is only safe when the professor runs the full analysis themselves beforehand. He avoids situations where hallucinations or inaccuracies could harm his use of AI so he tends to use it on specific easy tasks like extracting data. He described the responsible sequence for AI use as:
solid conceptual knowledge → a well-designed prompt → careful evaluation → then application in class.

We also discussed whether AI could replace teaching assistants. He explained that AI chatbots might be useful as support tools, essentially “course bots” that could answer routine questions about the course, provide access to lecture notes, or surface reminders. But he emphasized that these tools wouldn’t solve the real issue he’s been facing lately: students rarely attend office hours anymore or openly admit when they are confused. He described it as genuinely shocking how much the number of students coming to him has declined over the years.

When I suggested that students might avoid office hours because they’re afraid of looking like they lack understanding, and worry a professor might judge them, he dismissed that explanation immediately, saying his teaching style is completely non-judgmental. In his view, it’s not fear of the professor; it’s the broader culture of technology, comparison, and performance. Students today feel more pressure to appear competent at all times, and they hesitate to say “I don’t get it.” A course bot might even make this avoidance worse by giving students another way to avoid real conversations. 

He insisted that human TAs cannot be replaced because emotional intelligence, empathy, and nuanced real-time judgment can’t be automated. Still, it’s hard to ignore the underlying tension: even though he says humans are irreplaceable, it feels like everyone, including him, knows we’re living with a ticking clock. AI might not replace human interaction entirely, but the fear that it will eventually displace roles like TAs is becoming harder to deny. He even said that people rarely ask TAs for anything in his class so this proves my analogy.

When the conversation shifted to AI replacing accounting or finance jobs, such as internal auditing, financial modeling, or fraud detection, he called this assumption “too shallow.” Job loss is real, he said, but it has always happened with every technological shift. Old jobs disappear, and new jobs, ones we can’t currently imagine, emerge. Twenty years ago, no one predicted data scientists or media buyers. Likewise, he expects AI to create entirely new accounting and finance roles that require new skill sets, but the world is currently moving too fast to define them clearly. However, I feel that this is not true at all since AI might yes open new job opportunities for people, it will definitely close other opportunities for others. I tried mentioning this to him, but he just did not believe in this. 

Regarding internal auditing(which is something interesting to me since I was thinking of pursuing it), he acknowledged that AI can automate certain tasks in internal auditing like collecting certain margins like profit margins from financial statements, but training a model to replicate the judgment, experience, and tacit knowledge of an auditor is extremely difficult. Human auditors rely on an accumulation of experiences, industry exposure, and intuition that is hard to encode. Their analysis that people like CEOs rely on, he feels, can never be replicated by AI. He said this is “not impossible, nothing is impossible now, but very hard,” and pointed to scandals at major firms like PwC and Deloitte as examples of AI misuse or over-reliance. For example, the recent PwC incident where consultants were required to submit tax-related research to a government entity, and one consultant ended up using AI-generated “sources” that did not actually exist. The fabricated references created a major scandal and raised serious questions about oversight and accountability. To me, this case perfectly illustrates why AI cannot be blindly trusted or used indiscriminately across all fields. It highlights the danger of assuming AI output is automatically reliable, especially in areas where accuracy and credibility are essential. However, it still did not convince me of the fact that AI cannot actually replace internal auditors and his answers felt that they were coming from some bias (as he is the accounting professor). 

He also described practical ways AI helps him professionally. One of his main examples was extracting income statement data from PDF annual reports, especially in Egypt where financial statements are often poorly formatted. He frequently uses AI to convert messy PDFs into clean Excel files for analysis. When combined with a strong financial prompt and a good large language model, AI can generate:
• investor-focused reports,
• highlight company strengths,
• identify risks,
• prepare talking points for investor meetings,
• and suggest questions an investor might ask.

But he emphasized again that the usefulness depends entirely on the human’s financial knowledge and prompting skills. Without those, “AI will look stupid.” With the right inputs, however, it automates the time-consuming, mechanical parts, such as ratio calculations, while humans handle interpretation.

He also warned that AI cannot reliably create forecasts using only a large language model. Forecasting requires structured historical data and more advanced AI techniques such as retrieval-augmented generation (RAG), which is better at pulling information from specific documents and reducing hallucinations. He explained the difference between LLMs and RAG models using a metaphor:
• The LLM is like a brilliant person sitting in a huge library,
• The RAG model is like a specialist with a smaller set of curated books and better retrieval skills.
The RAG model “fine-tunes retrieval” and can answer more accurately using structured data sources, but both models still require strong human oversight.

He concluded by reminding me that AI is powerful but fragile. It automates routine steps, but interpretation, ethical judgment, and tacit knowledge remain human responsibilities. If someone uses AI without understanding Finance, they will harm themselves more than they benefit.

Step 4 Second Interview (Dr Nourhan)

I interviewed Dr. Nourhan, a Finance professor at AUC who teaches portfolio and investment-related courses, and she offered a very critical perspective on AI. She explained that although AI is already used in classrooms for quick summaries, concept explanations, and scanning research papers, she believes students are misusing it and becoming less engaged. When I asked about this in relation to insights I had gathered from Dr. Moataz, she agreed that office-hour attendance has noticeably dropped. She attributes this decline partly to students’ overreliance on AI, which she believes encourages avoidance rather than genuine learning.

When I asked her specifically about AI in portfolio management, she argued that AI often fails at calculations when prompts are poorly written and therefore cannot replace the human judgment required to interpret markets, understand investor psychology, or read uncertain macroeconomic signals. I challenged her by pointing out that humans also make mistakes, especially under pressure. She acknowledged this but emphasized that AI’s reliability remains inconsistent and that much more research is needed before trusting AI-driven models with people’s money. For her, the core issue is not whether AI can compute but whether it can understand, and she believes it cannot.

Regarding AI in education, she stated that AI’s adoption is inevitable but not inherently beneficial. She compared AI to Google, useful for retrieving information but never a substitute for reading, thinking, or developing analytical skills. She worries that students treat AI as a shortcut instead of a learning tool, and she personally avoids relying on it. When I asked how she uses AI herself, she said she limits it to simple tasks, such as summaries, and insists she “never and will never use it” for personal use, preferring Google instead. Her stance reflects a strong skepticism toward AI’s reliability.

On the topic of financial inclusion, she believes AI can support underserved groups through tools like prepaid cards, mobile wallets, and budgeting apps, but only if users are educated. Without digital literacy, AI could confuse or mislead people rather than empower them. She emphasized that AI is a tool, not a solution, and that the responsibility ultimately lies with the users. From her perspective, AI can enhance financial inclusion only if people know how to use it safely and intelligently.

Throughout the interview, she repeated that AI should remain a tool, not a teacher, and that human interaction and student responsibility are crucial for real learning. What stood out to me was how strongly she blamed the misuse of AI on students. She insisted it is entirely the student’s responsibility to use AI properly, read the outputs, and think critically before submitting work. While I understand the importance of student accountability, I found this perspective limited. Expecting students to use AI correctly without proper guidance seems unrealistic, and when I pushed her on this, she maintained that professors are not responsible for how students choose to use AI.

When I asked about the possibility of using an AI chatbot in the course, she strongly rejected the idea, arguing that AI hallucinates too often and cannot be trusted. Even when I challenged this by noting that AI improves every year, moving from simple text generation to document analysis, video creation, and even slide-making, she remained firm in her belief that AI will never be reliable enough for education. This led to the most difficult question: whether AI could ever replace professors and teachers. The question seemed sensitive; she paused before insisting that AI will never replace educators and that human interaction is irreplaceable. She argued that AI may replace mechanical tasks in Finance, like financial modeling or parts of portfolio management, but only for those who use AI “stupidly” without knowledge.

She compared the current AI wave to the rise of Google twenty years ago, saying that search engines never replaced teachers, and therefore AI will not replace them either. I found this argument unconvincing, because AI is not simply a search engine, it is far more sophisticated. While she saw AI as unreliable and static, I saw it as evolving rapidly. Her skepticism, while understandable, seemed to overlook the reality of how fast AI capabilities have grown.

Step 5 Reflection (My Honest Perspective)

After doing both interviews and reflecting on what I personally see happening around me, I realized that I feel very differently from what many professors believe. Everyone keeps saying that my generation is “lucky” because we grew up with AI and have access to tools previous generations never had. But the more I think about it, the more I feel the complete opposite. I don’t feel lucky at all. I feel like we are the generation that will have to work ten times harder just to survive in a world where AI can replace half the tasks we study years to master. Instead of feeling blessed, I feel anxious and almost rushed, like we’re racing against a technology that is evolving too fast for anyone to meaningfully control.

What surprised me most from the interviews was how strongly both professors insisted that AI cannot replace them or the human aspects of Finance. Dr. Moataz framed this fear of replacement as something “too shallow,” almost like a misunderstanding. But to me, nothing about this fear feels shallow. When I look at what AI is doing right now, pulling financial statements from PDFs, generating investor reports, creating clean ratio tables instantly, building models, analyzing risk patterns, those are tasks that juniors and interns used to spend hours on. Those were the tasks that gave new graduates experience. If AI is doing all of that, then what’s left for us in our first jobs? It’s hard for me to believe that new, imaginary future jobs will magically appear to save everyone. That hope sounds like hype, not evidence.

What also shocked me was how Dr. Nourhan blamed students entirely for the misuse of AI. She repeatedly said it’s the students’ responsibility to use AI correctly, think critically, and not become dependent. But how are students supposed to use AI “properly” when no one teaches them how? How can professors say they’re not responsible, yet complain about the outcomes? Her refusal to acknowledge that guidance is missing felt unfair. It made me realize how much of the AI conversation in education is built on blame rather than shared responsibility. The reality is that both students and professors are confused, overwhelmed, and trying to pretend they aren’t.

The part that scared me the most was the contradiction between what they say and what they fear. They kept insisting that AI cannot replace teachers, TAs, auditors, or portfolio managers because these jobs require “judgment” and “experience.” But neither of them could deny that AI is speeding up fast, learning to do things we once thought impossible. Only a few years ago, no one imagined AI making full models, interpreting PDFs, creating presentations, or mimicking reasoning. And now it’s happening. If something improves this quickly, how can anyone confidently say it will never replace us? I left both interviews feeling that their confidence came from personal bias or professional self-protection, not from real evidence.

Emotionally, this entire process made me feel uneasy, honestly. I don’t share the professors’ calmness. I don’t believe their optimism. I feel like we are heading toward a world where human value will be measured by how different we are from machines, not by how skilled we are. It terrifies me that AI could push capitalism into even harsher forms, where efficiency is everything, where fewer humans are needed, and where people who can’t keep up face unemployment, depression, or worse. I sometimes feel like we are walking straight into a future of intense competition, burnout, and maybe corruption, because once AI can manipulate predictions or information (which I believe it already can through biased data), the markets will become even more dangerous.

But this assignment also pushed me to see that fear alone isn’t useful. If AI will continue growing, then what I need to do now is learn how to verify, critique, and question its outputs. I need to become the kind of person who doesn’t just use AI, but understands its limits, catches its mistakes, and knows when something is too good to be true. The professors kept saying that judgment and interpretation are what humans bring, and maybe that’s the one thing we still have. I don’t think that will save everyone from being replaced, but it might be what protects me from blindly trusting something that can ruin decisions.

So my reflection ends in a strange mix of fear and determination. Fear because I know AI will reshape everything, probably in ways more destructive than the professors want to admit. Determination because I know I can’t run from it. I have to learn to stand next to it without letting it erase me. That means developing real financial knowledge, understanding AI beyond surface hype, and refusing to rely on it blindly. Maybe AI won’t kill human agency completely, but only if we fight to keep that agency alive.

Link of the interview audio part one and part 2, and transcript https://drive.google.com/drive/folders/13paSnIIT_F62aGoHt98r7ByJ0V6w8ee2?usp=sharing Prompt: https://chatgpt.com/share/692f28cd-fdc8-800c-b3a0-aa24c73b8cd3 I only recorded Dr Moataz, But Dr Nourhan just rejected being recorded I used Ai in fixing grammatical errors, but all of this was written by me It should be noted that I am not depressed but I am very realistic