In today’s Brainyacts:

1. AI wearables are coming fast

2. Ideas on building an AI practice area (riches in niches!)

3. Meta and your Facebook posts (🤔) and more AI model news

4. The 3.5 day work week and other related news

👋 to new subscribers!

To read previous editions, click here.

Lead Memo

👂🤖 Always On. Always Listening

The age of AI wearables is totally here . . . and it is awesome and scary!

Just think of all the legal work to come from this new AI category!!

Here are three wearables:

#1 Rewind Pendant

#2 Ai PIN

#3 TAB

It is worth watching this 8-minute demo on TAB. It is goofy at times but stay with the demo.

🚀 Unlock Your Legal Future with The Generative AI Blueprint 2.0 🚀

👋 Hey Trailblazers in Legal AI

📣 Big News! I’m turbocharging my Generative AI Blueprint course with Module 3: Advanced Skills! This course is more than just a tutorial. It's a strategic game-changer for your practice.

What's Cooking in Module 3:

• 🛠 Advanced Skills: A concentrated 40-minute blitz designed to make you a GenAI virtuoso.

• 🔥 Pragmatic Value: Not just theory, but applicable, battle-tested techniques.

• 🎯 Unprecedented Efficiency: Sharpen your toolset to tackle the most complex legal and business challenges with finesse.

⏳ Don't Be Left Behind: We're at the cusp of a paradigm shift in the legal industry. Are you going to watch it unfold, or lead the charge?

👇 Enroll Now and Get Ahead of the Curve: Generative AI Blueprint

Spotlight

AI Insurance & Its Promise of Growth for Legal Services

Two articles today I want to highlight and then present ideas on the impact on legal services. If you want to niche down in an AI-related practice, this is for you.

First Article

From the Wall Street Journal

Insurance companies are exploring the emerging field of AI and generative AI risk, offering financial protection against model failures, akin to the rise of cybersecurity insurance. Such policies aim to alleviate risk-management concerns for corporate technology leaders, CEOs, and legal departments. While it's still early, there is significant interest in AI insurance, with major carriers potentially offering specialized coverage for financial losses from AI-related issues such as cybersecurity risks, copyright infringement, and data leaks.

Companies like Munich Re and Toronto-based startup Armilla Assurance have already introduced AI insurance products. Munich Re covers financial losses if a company's in-house AI model errs in ways a human wouldn't, and Armilla Assurance offers a product warranty for AI models, backed by reinsurers like Swiss Re.

Tech giants Microsoft, IBM, and Adobe are also stepping in to manage risks. For instance, Microsoft has committed to defend and pay for lawsuits resulting from the use of its generative-AI-based Copilot tools, provided users follow built-in guardrails.

Analysts expect that insurance policies covering AI risks will become commonplace in the coming years, given the estimated trillions that generative AI could add to global economic output.

However, insurance should not be seen as a substitute for good risk management; it's crucial for companies to have their own safeguards and security measures in place as a first line of defense. The evolving nature of AI technology also poses challenges for insurers in terms of risk assessment, requiring them to adapt and update their methodologies continuously.

Second Article

From a recently published paper.

A research team from the University of North Carolina, Chapel Hill has unveiled compelling findings on the challenges of eliminating sensitive information from large language models (LLMs) like ChatGPT and Google's Bard. Their research highlights that while it may be technically feasible to "delete" specific data from these models, verifying the effective removal of such data is a complex task. This complexity stems from the architecture of LLMs, which are trained on vast databases and later fine-tuned, effectively solidifying the information into a "black box" of unidentifiable weights and parameters.

Given the inability to retroactively delete particular files from a trained model, developers implement "guardrails," using techniques like reinforcement learning from human feedback (RLHF) to discourage the model from producing harmful or sensitive information. Yet, this approach is reactive and incomplete; it can't guarantee the removal of sensitive data from the model's memory.

In their experiments, the researchers found that even state-of-the-art editing techniques fail to fully delete factual information. Using GPT-J as their test subject—a significantly smaller model than GPT-3.5—they found that sensitive facts could still be extracted up to 38% of the time by whitebox attacks and 29% of the time by blackbox attacks. They also managed to develop new defense methods, but acknowledge that these defensive measures are perpetually in a game of "catch-up" with new offensive techniques.

Implications for legal services growth

Reading these articles got me thinking of all the potential areas for lawyers to build new practices focused on. This is just a random brainstorm of ideas but it took almost no time at all. Spending more time unpacking and exploring these will certainly yield significant upside.

I personally have been considering whether I could start a virtual practice (licensed in DC) that focuses on my interest and budding expertise in certain areas of generative AI. I welcome your ideas!

Ok here is a starter list:

1. Legal Advisory on AI Risk Management:
Given the emerging field of AI insurance, businesses are grappling with understanding the full range of risks associated with AI. As a legal expert, you can carve out a niche by specializing in AI risk assessment, governance, and compliance.

2. AI Contractual Frameworks:
Considering IBM and Microsoft are adjusting standard contractual obligations to cover generative AI models, you might specialize in drafting and negotiating these new-age contracts that account for AI-specific liabilities and defenses.

3. Legal Audits for AI Insurance Eligibility:
As AI insurance providers like Munich Re and Armilla Assurance develop frameworks for assessing risk, offering a service that prepares companies for these audits could be valuable. Ensure that companies' AI models, training data, and use-cases align with the underwriting requirements of insurance providers.

4. Copyright and Data Privacy:
Generative AI can inadvertently infringe copyrights or leak data, even in your own enterprise AI model. Legal solutions for preventive measures, indemnification clauses, and defense in case of such events would be an asset to clients.

5. Policy Advocacy and Standard Setting:
Insurance and technology companies are early into understanding what AI governance should look like. There's an opportunity to be part of the dialogue that sets these standards, thereby having a say in industry and potentially global policy.

6. Litigation and Arbitration Specialization:
With the introduction of AI, traditional legal skirmishes are now becoming more complex. Developing expertise in litigating cases around AI failures, insurance claims, and breaches would place you ahead of the curve. Needless to say, working with fact-finders on the massive and overwhelming deepfake problem represents significant opportunity.

AI Model Notables

• Meta's new AI assistant trained on public Facebook and Instagram posts

• Apple bulking up by hiring UK AI staff

• How to Stop Google Bard From Storing Your Data and Location

• Controversial AI Company Mistral Releases Chatbot Without Ethical Filters

News You Can Use:

➭ JPMorgan’s Dimon Predicts 3.5-Day Work Week for Next Generation Thanks to AI.

➭ Artificial Intelligence (AI) and the Practice of Law (The Sedona Conference Journal) Paper here. A review of it here.

➭ New data shows that without intervention, generative AI jobs will continue to cluster in the same big tech hubs.

➭ AGI House: A different kind of Silicon Valley accelerator (unsure what to call it but I like it).

➭ Tom Hanks is trying to defeat his deep fake that sells dental plans.

Who is the author, Josh Kubicki?

Some of you know me. Others do not. Here is a short intro. I am a lawyer, entrepreneur, and teacher. I have transformed legal practices and built multi-million dollar businesses. Not a theorist, I am an applied researcher and former Chief Strategy Officer, recognized by Fast Company and Bloomberg Law for my unique work. Through this newsletter, I offer you pragmatic insights into leveraging AI to inform and improve your daily life in legal services.