I had a great time on Friday! Looking forward to the October and November meetings.
Tips & Insights for Professors, Curious Teachers, and Motivated Students
For our next panelist spotlight, we would like to introduce you to Perry Binder. Perry
is part of a panel of four GSU faculty who will share their real-world
experiences, challenges, and strategies for integrating AI ethically and
effectively into their teaching practice as a part of our new fall
series, AI x AI.
Perry
Binder, J.D. is a Clinical Associate Professor of Legal Studies at
GSU’s Robinson College of Business. While practicing law in South
Florida, he got the teaching bug when he taught part-time at Miami Dade
College and Broward College. Perry’s latest book, Innovative College
Teaching (2024), has a chapter on “Using AI to Enhance Classroom
Activities and Student Projects” and offers teaching tips and insights
from a dozen current and former GSU master teachers. The book debuted as
an Amazon #1 New Release and #1 Best Seller (Teacher & Student
Mentoring) and received the International Book Award (Education &
Academic category). Each year from 2016-2022, Perry’s book, 99
Motivators for College Success, was sent to hundreds of rising high
school seniors nationwide as part of the Book Award Program at Randolph
College in Virginia.
Join
us for “The Trailblazers: GSU Faculty Leaders in AI Teaching
Innovation” virtually on Wednesday, Sept.17 from noon to 1 p.m.For our next panelist spotlight, we would like to introduce you to Perry Binder.
First Prompt: consumer question: My apartment complex had 10 cars broken into 3 months ago and 15 cars this month. no security has been added. If my car is broken into tonight do I have have a good lawsuit against the owners of my apartment complex?
Second prompt: I cannot afford an attorney -- please put into plain English what I can do now and will I have a good lawsuit
GPT response to second prompt
Maybe, but it’s not a sure thing.
The apartment owner only has to take “reasonable steps” to keep the property safe.
If crimes keep happening and they do nothing (no lights, cameras, patrols, notices), a judge might say they failed their duty.
But the law does not make landlords automatically responsible every time a tenant’s car gets broken into.
So: you might have a case, but it’s hard to win without proof that management knew about the crime problem and ignored it.
Write to management in email or letter: Tell them you’re concerned about repeated break-ins and ask what steps they will take (lights, cameras, patrols). Keep a copy. This shows they were on notice.
Document everything: Save police reports, photos, and any messages you get from neighbors about break-ins.
File a police report if your car is ever broken into. Keep the report.
Check your renter’s or auto insurance: Often this is the quickest way to get compensated, even if it feels unfair.
Legal aid: If you can’t afford a lawyer, look up free tenant legal aid groups or housing legal clinics in your area. They often help with exactly these situations.
If your car is broken into:
You can try to sue, but you’ll need to show the landlord knew there was a serious problem and didn’t act reasonably.
It’s not guaranteed money, but documenting now will make your case stronger later.
Using GPT-5 in “Study Mode” (August 25, 2025) Perry Binder
My Prompt: Let's go into study mode so I can learn about comparative negligence in Georgia tort cases
ChatGPT said:
Perfect — let’s switch into Study
Mode: Comparative Negligence in Georgia Tort Law.
Here’s how I’ll guide you:
1. Quick Explanation
In Georgia, comparative negligence is the rule that a plaintiff can still recover damages in a tort case even if they were partially at fault — but only up to a point.
Georgia follows a modified comparative negligence, 50% bar rule:
2. Key Points in Georgia
3. Quick Practice Hypothetical
Carla is hit by Dan while crossing the street outside a crosswalk. The jury finds:
👉 Question: How much does Carla recover under Georgia law?
Do you want to try solving this one, and then I’ll walk through the answer with you?
Perry Binder will be a part of a panel of experts at an online CETLOE event on Sept. 17, The Trailblazers: GSU Faculty Leaders in AI Teaching Innovation (learn more and register at link). Perry’s book, Innovative College Teaching: Tips & Insights from 14 Master Teachers, includes information about teaching with AI.
(only for GSU professors and staff)
Over the past year, I've made 20 podcasts with Notebook LM. For example, I created podcasts on pivotal court cases, such as copyright issues with AI, and two of my academic law papers.
Recently, LM generated a 7-min. MP4 from my two-page PDF conference handout on Embracing AI to Flip College Classes.
LM now allows you to upload various items, not just PDFs. Also, it is not just audio - LM turned my handout into a video (took about 20+ minutes to generate) with PPT-like slides that you can edit.
The LM paid version lets users control how long each audio/video should be — an important feature because most of my past podcasts are too wordy.
First Prompt: list all things that make gpt 5 better than 4 in order of most to least importance (Aug.15, 2025)
Results:
Here’s a structured list of the main improvements that make GPT-5 stronger than GPT-4, ordered from most to least important in terms of impact on real-world use:
Reasoning & Problem-Solving Ability – Handles complex, multi-step reasoning tasks more reliably and with fewer errors.
Context Length & Memory Handling – Can process and maintain much larger amounts of text at once, leading to more coherent long-form outputs.
Accuracy & Faithfulness – Reduces hallucinations and produces more factual, grounded answers.
Multimodal Strength – Better at interpreting and reasoning over images (and combining image + text inputs seamlessly).
Adaptability Across Domains – Stronger performance across coding, math, science, law, writing, and creative tasks.
Speed & Latency – Faster responses with more efficient token usage.
Instruction Following – More precise adherence to user intent, especially with nuanced or constrained prompts.
Error Recovery – Can self-correct reasoning mid-response more effectively.
Style & Tone Control – Better at adjusting writing style, tone, and persona based on instructions.
Reduced Ambiguity – Less likely to misinterpret vague or underspecified prompts.
Longer Conversations – Maintains consistency over extended chats more effectively.
Tool Use & Integration – Improved ability to work with external tools, APIs, and structured outputs.
Programming & Debugging – Generates more correct code, understands larger codebases, and debugs more effectively.
Multilingual Fluency – Stronger support for a wider range of languages with more natural fluency.
Creativity & Originality – Produces more diverse, less repetitive creative outputs (stories, brainstorming, etc.).
--------------
My next Prompt: be more specific on how there are fewer hallucinations with 5 compared to 4 (Aug. 15, 2025)
Results:
Training Data Improvements
GPT-5 is trained on a larger and more carefully curated dataset.
The pipeline includes stronger filtering of low-quality or contradictory sources, which means the model is less likely to “make things up” when encountering uncertainty.
Better Alignment & Post-Training
Reinforcement Learning with Human Feedback (RLHF) and newer alignment techniques are more fine-tuned in GPT-5.
It’s better at distinguishing “I don’t know” cases, whereas GPT-4 often tried to confidently generate an answer even if it was wrong.
Improved Internal Verification
GPT-5 uses enhanced internal consistency checks during generation.
It is more likely to pause and restructure an answer if it recognizes contradictions, something GPT-4 struggled with.
Fact-Grounding & Retrieval Hooks
GPT-5 is more compatible with retrieval-augmented generation (RAG) pipelines and citation mechanisms.
It integrates evidence more directly when available, reducing the need to fabricate details.
Domain-Specific Accuracy Gains
Coding: GPT-5 produces significantly fewer “phantom functions” or nonexistent libraries compared to GPT-4.
Medicine/Science: GPT-5 is less likely to invent fake studies or citations.
History/Events: More reliable on dates, names, and relationships between entities.
User Feedback Loops
GPT-5 is better at self-correcting mid-conversation if the user challenges an answer. GPT-4 often doubled down instead of revising.
📊 Empirical Example (simplified):
In internal benchmark tests:
GPT-4 hallucinated ~15–20% of the time on open-domain fact questions.
GPT-5 cut that to around 5–10%, depending on the domain.
Very kind review:
Reviewed in the United States on July 31, 2025Honored to get this review:
Reviewed in Spain on July 30, 2025
Short YouTube Clips on... Shy Student & Book Giveaway / College is a Marathon Story / Welcome to College Moment / Social...