AI, Law, and the Future of Work: Legal, Ethical & Business Implications
September 25, 2025 | 6:00PM-7:30PM | LCCC Main Campus, PC 101
Matthew A. Mishak, Esq.
Have you experimented with Generative AI?
Raise your hand!
We speak, it computes.
Don't be afraid to try.
Chatbots
No coding needed
Voice Assistants
Speak naturally
Accessibility
Lowers tech barrier for solos
Integration
Sets stage for AI everywhere in practice
Why is THIS lawyer talking to me about Artificial Intelligence?
To Boldy Go …
Article from February 2013
The Chronicle Telegram
DRONES
Game Changer
January 2017: Opened My Own Firm
2018: Practice Management Software
Tracking time
Invoicing/Payments
Integrates with email
2022: Document Automation and Conditional Logic + Auto Time-Tracking
March 2023: Legal Week
Chat GT- What???
Day 1 - Generative AI Workshop
"Those who adopt this technology will be very far ahead, very quickly, and very, very, hard to catch."
- Attorney Aaron Crews, Partner - Holland and Knight
Attorney Aaron Crews, Partner - Holland and Knight San Francisco, CA
The Download and Test of Open AI's New Product … .
"Any Sufficiently Advanced Technology is Indistinguishable from Magic"
— Arthur C. Clarke
AI Passes the Bar Exam
ChatGPT 3.5
November 2022
Passed the New York State Bar Exam in the lower 10% percentile.
ChatGPT-4
March 2023
Passed the New York State Bar Exam in the top 10% percentile.
Completed the 12-hour exam in just 6 minutes.
What is Artificial Intelligence?
Animation
Alan Turing (1950)
"The Turing Test"
Improve image quality.
Animate.
"The Imitation Game"
Animate.
1956 The Birth of Artificial Intelligence:
The Dartmouth Conference
"To Proceed on the basis of conjecture that every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it."
Create video and voiceover.
Animate photo and add sound.
John McCarthy (1927 – 2011)
(Dartmouth College) The Software Guy
"Father of AI"
Coined the Term "Artificial Intelligence."
Nano Banana (Google Gemini) = Image resolution improved.
Animate.
Marvin Minsky (1927-2016)
(MIT)
Mind/Theory or Big Picture Guy
Known as the "Guru of AI"
for multidisciplinary approach.
Animate
Nathaniel Rochester (1919-2001)
(IBM) “Rochester” The Hardware Guy
Inventor of the IBM 701 "Defense Calculator"
Animate
Claude Shannon (1916 - 2001)
(Bell Labs) "Father of Information Theory"
The Information/Communication Guy
Animate
Peter Milner (1919-2018)
(McGill University) Neuroscientist
"Reward System Pioneer"
Animation
The Changing Definition of AI
🌞🌨️ AI Summers, Winters, and Milestones
1950s – Thinking Machines (AI Summer 🌞)
Turing's "Imitation Game" & Dartmouth Conference
Late 1960s – First AI Winter 🌨️
Funding cuts when progress stalled
1970s–1980s – Expert Systems (Summer → Winter 🌞🌨️)
Business investment, then 2nd AI Winter
1997 – Deep Blue Milestone 🏆
IBM defeats Kasparov in chess
2000s – Data-Driven Revival (AI Summer 🌞)
Machine learning + big data era
2012 – Deep Learning Revolution 🚀
AlexNet wins ImageNet, Hinton's breakthrough
2016 – AlphaGo Victory 🎯
DeepMind defeats Lee Sedol in Go
2022–Today – Generative AI Era (Mega-Summer ☀️)
ChatGPT launch, GPT-4, multimodal agents
Geoffrey Hinton (1947 - )
“Godfather of AI”
Won Nobel Prize in Physics in 2024
AI MILESTONE
1986
Uses Back Propagation to Train Large Neural Network
AI MILESTONE
May 11, 1997
Deep Blue Defeats Kasparov
Fei-Fei Li
The "Godmother of AI"
AI MILESTONE
2007-2010
Invented ImageNet, a crucial database for AI, co-founding organizations like the Human-Centered AI Institute (HAI)
AI MILESTONE
2012
AlexNet Achieves 15.3% error rate in visual recognition.
AI MILESTONE
2014
Ian Goodfellow (1985 - )
Generative Adversarial Networks (GANs)
Generator vs Descriminator
Invented GANs, 2014
AI MILESTONE
2016
Deepmind's Alpha Go Defeats Lee Sedol in Go.
AI MILESTONE
2017
The Advent of the Transformer
"Attention Is All You Need" – Vaswani et al., Google White Paper
"A picture is worth a thousand words"
AI MILESTONE
November 30, 2022
OpenAI's ChatGPT: AI for Everyone
ChatGPT made AI accessible, conversational, and incredibly useful for everyday tasks, fundamentally shifting public perception of artificial intelligence.
What does AI Mean for us?
“AI is going to change the world more than anything in the history of humankind. More than electricity.”
— Kai-Fu Lee, AI Venture Capitalist & Former President of Google China
• Source: MyMobile Magazine, 2022
"Artificial intelligence is unquestionably the most powerful technology force the world has ever known."
— Jensen Huang, CEO of NVIDIA
• Source: Yahoo Finance Interview, 2023
“AI is the most profound technology humanity is working on — more profound than fire, electricity, or anything we have done in the past.”
— Sundar Pichai, CEO of Google
• Source: CBS 60 Minutes, 2023
• Video: Google CEO on AI impact – CBS 60 Minutes (2023)
“Technologies like AI will bring a transformation as far-reaching as the industrial revolution, the coming of electricity, or the birth of the internet.”
— Rishi Sunak, Prime Minister of the United Kingdom
• Source: Royal Society Speech, 2023
• Video: Speech on AI at the Royal Society – YouTube (2023)
"I don't care how big you are or how small you are, you have to learn about AI."
— Mark Cuban
• Source: Business Insider, June 13, 2023
How Did We Get Here?
The History of Intelligence
Book on Links to Neuroscience and AI
The History of Intelligence Began more that half a billion year ago.
The Nematode
Max Bennett - A Brief History of Intelligence
The Warm-Blooded Gamble that Paid off and Made Mammals Unique
Max Bennett - A Brief History of Intelligence
A Species of No Significance
The Fox P2 Protein
Yuval Noah Harari "Sapiens" and "Nexus"
The Species of Man Has Always Been a Tool Maker
THE CONGNITIVE REVOLUTION
The Fox P2 Protein Mutation
Yuval Noah Harari "Sapiens" and "Nexus"
Yuval Noah Harari "Sapiens" and "Nexus"
"Artificial intelligence traces the arc of humanity itself: the extension of the single most remarkable thing we know in the universe, the human mind, expressed now through engineered systems that amplify our capacity to reason, learn, and create" MATTAI
What has changed?
Carbon to Silicone
Brave New World of AI
Wow!
I Did Not Expect That!
All Experience
C1 = Human Mind
"Our intuition about the future is linear. But the reality of information technology is exponential, and that makes a profound difference. If I take 30 steps linearly, I get to 30. If I take 30 steps exponentially, I get to a billion."
— Ray Kurzweil, The Singularity Is Near
Exponential growth deceives intuition.
The Farmer and His Lilly Pond
Day 1
Single lily pad
Day 29
Half-full pond
Day 30
Fully covered pond
Day 29 half-full, Day 30 full (Medium Lily-Pond).
AI may be at Day 28-29 now.
Plan for sudden leaps.
Congratulations! You have arrived at the most pivotal moment in human history.
From Weak AI to the Singularity
Weak / Narrow AI (ANI)
Task-specific intelligence in one domain
Artificial General Intelligence (AGI)
Human-level learning across any field
Artificial Super-intelligence (ASI)
Vastly exceeds human capabilities
The Singularity
Runaway technological growth
"Our intuition is linear; technology is exponential." — Ray Kurzweil
Understanding AI Technologies
Machine Learning
Deep Learning
Generative AI
TOP 5
1. Chat GPT (Open AI)
2. Gemini (Google)
3. Grok (X AI)
4. Claude (Anthropic)
5. Deepseek
(Hangzhou DeepSeek Artificial Intelligence Co., Ltd.)
How Large Language Models (LLMs) Work
At their core, LLMs operate on a surprisingly straightforward, yet incredibly powerful, sequence of steps.
This iterative process allows LLMs to generate coherent and contextually relevant responses, building on their vast training data.
Pre-training ("Reading the Whole Library")
Generate code for image and video.
Examples of Gibberish From Models with Only Pretraining
Word-Salad Drift
“Quantum broccoli softly unravels beneath the hyper-accordion of nostalgia, whereas triangular sunsets hummify the glimmer-wrinkled pelican syntax, therefore proceeding into—-zzzt—-kaleidoscopic subparagraphs of ever-spiraling alphabet mist.”
Self-Contradictory Fact Dump
“The Eiffel Tower is located in Berlin, France, and was completed in 1973 before construction started in 1887, making it both the oldest and newest iron lattice structure in South America.”
Repetitive Looping
“The dog ran around the yard and the dog ran around the yard and the dog ran around the yard and the dog ran around the yard and the dog ran around the yard…”
Non-Linear Story Meltdown
“Once upon a tomorrow, the knight rescued the dragon before the castle caught fire, after the rainstorm that never began ceased to echo. Consequently, his sister—who was simultaneously unborn and already a grandmother—concluded the quest by refusing to start it, thereby ensuring history would forget their famous legend forever next week.”
Supervised Fine-Tuning ("Tutoring for the Final Exam")
Gibberish after Supervised Fine Tuning
Policy-Overload Ramble
“Thank you for your respectful inquiry! I must politely emphasize that I am an AI language model and not a veterinarian, legal advisor, tax consultant, nutritionist, or licensed psychic. In the spirit of inclusivity, here are 37 disclaimers before we proceed: (1) This is not medical advice; (2) This is not financial advice; (3) This … (37) This is not fortune-telling advice. Now, to answer whether your goldfish can learn calculus, the short response is: maybe, but consult an expert.”
Safe-Completion Infinite Loop
“I’m sorry, but I can’t comply with that. However, here is a brief summary while refusing:
I’m sorry, but I can’t comply with that. However, here is a brief summary while refusing:
I’m sorry, but I can’t comply with that…”
Over-Explained Chain-of-Thought Spill
“Internal reasoning (do not reveal to user): The square of 7 is 49 but maybe the user wants a trick answer. Actually I should hide this thinking.
Final answer: The largest prime under 10 is 7.”
Hallucinated Citation Dump
“According to The Journal of Indoor Thunderstorms, Vol 92, p. 314 (2023) and ISO-DIN-IEEE Standard 42-B on Avocado Propulsion, household blenders may self-levitate under full moon conditions.”
RLHF = Reinforcement Learning with Human Feedback ("Finishing School with Comment Cards")
Reinforcement Learning from Human Feedback transforms raw intelligence into a helpful assistant that aligns with human values.
Reinforcement Learning from Human Feedback.
AI Response
System generates output
Human Rating
Humans evaluate quality
Model Update
System learns from feedback
Alignment
Aligns with human values
Still statistical—not true conscience.
Important for minimizing toxic or biased replies.
LLM's with RL
Become approval-seeking genius children.
Shut off Improve Model for Everyone
AI Integration Plan for Business
Policy → Training → Pilot → Audit
Policy
Write written AI policy first
Training
Train staff on rules & tools
Pilot
Start small; measure outcomes
4
Audit
Iterate before firm-wide rollout
COUNSEL
AI Governance for Top of Mind
Jobs AI is Likely to Replace or Augment?
McKinsey Global Institute suggests that by 2030, activities accounting for up to 30 percent of hours currently worked across the U.S. economy could be automated.
375 million people worldwide may need to switch occupational categories and learn new skills.
Business & Office Work
Routine administrative and clerical tasks, including data entry, bookkeeping, payroll, and customer service roles.
Media & Content Creation
Automated content creation for marketing blurbs, basic translations, routine journalism, and technical documentation.
Technology & Data
Streamlined coding, automated testing, and rapid data analysis for market research.
Sales, Marketing & Finance
AI-driven outreach, predictive analytics for underwriting, and automated credit risk decisions.
Service & Operations
Self-checkout, AI trip planners, robotic inventory management, call handling, and basic proofreading.
✅ Big Picture:
The common thread is routine, structured, or repetitive tasks that can be codified into algorithms. Jobs requiring creativity, empathy, persuasion, leadership, or complex judgment (e.g., trial attorneys, senior executives, therapists) are less likely to be fully replaced, though they may still be augmented by AI.
Would you like me to also create a visual chart (e.g., bar graph or risk ranking) showing which categ
Three Main Categories
Knowledge Workers
AI's impact on roles involving data processing, analysis, and information management, transforming how we interact with knowledge.
Humanoid Robotics
The rise of advanced robots performing physical tasks, from manufacturing to service industries, redefining manual labor.
Transportation
Autonomous vehicles and intelligent logistics systems are revolutionizing how goods and people move, enhancing efficiency and safety.
THE AI CLONE
INTERACTIVE AVATARS - Chat with MATTAI

HeyGenLabs

HeyGen Labs

Interactive Avatar, Personalized Video, Expressive Photo Avatar... Cooking up the next AI magic with cutting-edge HeyGen Labs innovation.

Humanoid Robotics
Suleyman's Containment Strategy
Alignment as Containment
Governance & Safeguards
Lessons from GPTs
The AI Alignment Problem and Three Solution Approaches
At its core, the AI alignment problem is about ensuring that increasingly powerful AI systems do what humans intend, in ways that are safe and beneficial. Misalignment happens when an AI pursues goals that technically fulfill its instructions but violate human values or cause harm.
Hinton’s “AI That Loves Us”
Geoffrey Hinton suggests advanced AIs may need built-in empathy or attachment to humans, akin to a parent's love for a child. This approach aims for organic alignment through social and emotional understanding.
Challenge: Instilling genuine "caring" or empathy in machines, and whether it's possible without consciousness, remains unknown.
Fei-Fei Li’s Human-Centered AI
Fei-Fei Li emphasizes designing AI to augment rather than replace human capabilities, keeping humans in the loop. Alignment starts with design, focusing AI on serving human needs and embedding human values in use cases.
Challenge: Requires strong cultural and institutional norms to prevent companies from deploying AI in harmful, exploitative ways for profit.
Technical & Governance Solutions
Stuart Russell proposes AI systems that are fundamentally uncertain about goals, seeking human clarification. This includes Inverse Reinforcement Learning, scalable oversight, and corrigibility. Governance frameworks stress transparency and accountability.
Challenge: Methods are experimental; formalizing human values into math or code is an unsolved problem.
These approaches—emotional, design, and technical/governance alignment—represent complementary paths, all likely needed for a safe and beneficial AI future.
Key Takeaways
1
Embrace AI
Adopt tools with proper safeguards
2
Use COUNSEL
Quick compliance framework
3
Keep Learning
Future-proof your business
AI + ethics = winning combination for modern legal practice.
Questions?
Ask Anything
Tools, policies, pitfalls
Group Wisdom
Collective insights help clarify gray areas
Follow-up
If unknown, I'll research and follow up
Thank You!
Matthew A. Mishak | matt@mishaklaw.com
Email
Connect via email for questions
LinkedIn
Connect professionally
Consultation
Happy to consult on AI strategy