Is AI the ultimate lawyer?

Should machines make life-changing decisions?

All Rise!

The Honorable Artificial Intelligence, presiding.

Please be seated.

The topic du jour is AI in the legal industry and how it may affect business, work, and our lives.

Today’s headlines:

  • AI is entering the legal industry. Should we be concerned or excited?

    • ChatGPT for legal help

    • ChatGPT for pro bono work

    • Using ChatGPT to prepare testimonials and arguments in Civil court

    • Harvey AI - the world’s first AI for law firms

    • DoNotPay - consumer rights-focused startup

    • AI and privacy concerns

    • Wrapping up

  • AI art of the Week: Women’s handbags

  • Useful AI tools for your business or work (new section)

  • AI experiment of the week

  • The latest robotics finds

  • Quick bites - interesting AI news

  • An ask

AI is entering the legal industry. Should we be concerned or excited?

There are multiple ways we can look at the applications and implications of AI on the legal industry.

In this deep dive, I will look at how it will affect the legal industry, as well as how it can affect businesses and you personally.

As a European, I hardly worry about the legal implications of anything I do or say as long as I stay within the confines of the law.

The same goes for my businesses.

It does not mean that I never interact with lawyers.

I do.

I need contracts or legal advice from time to time.

But if I were American or based out of the USA, I would be considerably more concerned about “legal matters.”

In Europe, we don’t take anyone to court if they call a man a he, instead of she/they/them/wolf/trumpet/spidermouse (looking at you, California).

Despite having the most lawyers in the world per capita, as much as 80% of Americans cannot afford legal help.

Armed with super powerful tools like ChatGPT and GPT-3/GPT-4 from OpenAI, would it be fair to say that the legal industry is about to be disrupted?

Let’s find out!

ChatGPT for legal help

It feels like ChatGPT is a household name by now.

I am biased, as I live in an AI bubble, and everyone talks about AI with me.

However, ChatGPT has proved to be a powerful tool for all kinds of situations, despite its biases and sometimes being flat-out wrong or making up stuff as it goes.

In December last year, I wrote an article on my blog about using ChatGPT to draft rental contracts - one commercial lease contract for New Jersey and one for a long-term rental in Texas (both fictitious, of course).

I am no lawyer and cannot say if the output I got was sound or not. But the mere fact that I, as your average Joe, can generate a draft contract in seconds represents an enormous shift in power.

What we used to pay a lawyer thousands of dollars can now be done for free in less than a minute.

The output cannot be trusted 100%, but it may cut legal costs for many people and businesses.

In the first issue of the Practical AI newsletter, I wrote about the panicking lawyer on Twitter, basically saying that the legal industry is “doomed” due to ChatGTP.

He showed how he could draft contracts lawyers generally charge a lot of money for in seconds.

If you have not seen it before, here it is:

But it is not only generating contracts we can do with ChatGPT and AI.

You can use it to review contracts as well. You can ask ChatGPT to summarize it and even ask if any points in the agreement are skewed in your disfavor. I wrote an article about reviewing legal contracts using ChatGPT here.

You should have a lawyer look over important contracts. But in my business, getting a contract summarized and reviewed by ChatGPT is usually good enough.

ChatGPT for pro bono work

In America, civil legal aid is not guaranteed under federal law.

Some people are lucky and get legal aid via public interest law firms, community legal clinics (pro bono), federally-funded legal services, pro bono lawyers, and private volunteers.

Imagine the impact ChatGPT will have on productivity for a pro bono legal clinic using ChatGPT.

They can help more people with their newfound gains!

That is, in my opinion, quite revolutionary and disruptive!

Using ChatGPT to prepare testimonials and arguments in Civil court

My friend Christopher runs one of the largest Facebook groups on ChatGPT called ChatGPT for Business & Life (free to join).

Back in January, he was sharing with the group how he used ChatGPT in legal contexts and that he was using it to help him prepare for Civil court.

The screenshot below shows one of his posts in the Facebook group.

The mentioned case in Civil court went well, and he came out with a positive result.

It is essential to mention that this should not be considered legal advice, and ChatGPT does not replace a real lawyer. Everything here is for entertainment purposes only. Use it at your own risk. I advise you always to check any legal matters with a lawyer.

Harvey AI - the world’s first AI for law firms

Legal startup Harvey AI made the news last week as Allen & Overy (A&O), the twelfth largest law firm in the world, announced that they had integrated Harvey AI into its global practice.

Harvey AI works with OpenAIs GTP-3, which is basically the same AI powering ChatGPT.

Harvey AI will be used by over 3,500 A&O lawyers across 43 offers in multiple languages.

During the trial period, A&O’s lawyers have done over 40,000 queries to Harvey. Today, about 25% of the company’s lawyers use Harvey daily, while about 80% use it once a month or more.

Integrating AI into legal practice is not straightforward and comes with constraints and needs to be monitored heavily. It also faces privacy concerns, in addition to AIs' tendency to “hallucinate” (make stuff up).

In the case of A&O, when a user logs into the Harvey portal, they are confronted by a list of rules for using the tool, the most important “You must validate everything coming out of the system. You have to check everything.”

Allen & Overy has chosen to not use personal data for its deployment of Harvey AI to avoid privacy pitfalls.

Lilian Edwards, professor of law, innovation, and society at Newcastle University in the UK, told Wired that “legal applications such as contract, conveyancing, or license generation are relatively safe areas to employ ChatGPT and its cousins (..) because law firms can draw on large amounts of highly standardized templated and precedents banks to scaffold document generation…”.

I think any business area dealing with a lot of standardized or structured data are excellent fields for AI to work its magic.

The legal sector, in that regard, seems a great place to deploy AI tools to assist legal professionals.


Harvey AI was founded in 2022 by Winston Weinberg and Gabriel Pereyra. It is backed by OpenAI’s Startup Fund, which invests in early-stage AI companies tackling major problems.

DoNotPay - consumer rights-focused startup

This is another company I had mentioned in a previous newsletter when they announced that they would fight a speeding ticket case in court using their GPT-3 powered AI.

Although electronic devices are banned in the courtroom, they planned to use hearing accessibility standards as a “loophole,” allowing their client to wear AirPods during the trial. This way, they would have a way to communicate in and out of the courtroom to give questions and receive answers from the AI.

Eventually, Joshua Browder, the creator and CEO of DoNotPay, had to cancel using the robot lawyer/AI in court, as he was threatened with six months in jail if he followed through with his plans.

DoNotPay continues to use its AI for legal work, although they will not be taking the AI lawyer into a courtroom anytime soon.

AI and privacy concerns

There are many concerns about our privacy in an AI-enabled world.

In a legal setting, these concerns are even more prominent.

While the outputs from legal AI will need careful monitoring, the inputs could be challenging to manage as well. How can one secure that the data submitted to an AI does not become part of the data model or its training?

Such a situation will likely violate confidentiality obligations, data protection, and privacy rights.

But isn’t this what we already do with ChatGPT?

In Europe, we have the General Data Protection Regulation (GDPR), which governs how much data about individuals can be collected and processed by companies.

In addition, the EU is stepping up its legislation around feeding generative AI tools with personal data. The proposed EU AI Act aims to regulate the use of artificial intelligence within its borders.

How does an AI navigate the legal framework in Europe?

Earlier this week, we could read about privacy concerns in Australia as the largest supermarket chain Woolworths expands the use of camera and AI technology at checkouts to over 250 stores.

Recently Amazon, Accenture, and now JP Morgan Chase banned the use of ChatGPT in their organizations due to privacy concerns.

Wrapping up

I try to be the guy who sees the glass half full.

Many people do not have access to legal help, and AI makes me optimistic on their behalf.

I believe that tools like ChatGPT can give David superpowers in the fight against the Goliaths of the world.

From a business perspective, I think ChatGPT can be particularly helpful for small businesses in helping them cut legal costs, understand what contracts actually mean, and how they may be skewed to benefit the other party.

Regarding specialized legal AIs like Harvey AI, I think it is a great tool that can help a service business create efficiencies where it previously was impossible to offer automation and productivity gains.

Harvey AI will create savings for law firms, meaning they can take on more business without hiring more staff.

Alternatively, it may also mean that the legal team may be able to go home earlier and spend more time with their families… (glass half full, remember…)

Overall, I am very optimistic about the future of AI in a legal context, and I am excited to see how it will all pan out.

AI art of the week - Women's handbags

Image created using generative AI Lexica.

I did not make today’s image. It was created by my friend Zeng, an AI art enthusiast, a fellow AI newsletter writer (WhoWhatWhyAI), and the creator of @femmestock.

The reason why I used her image today, vs. my own, is that after a lot positive feedback on her virtual handbags, she is planning to turn the bags into real-life products and sell them.

How cool is that?

From an idea to an AI concept to physical products.

We will see many more artists and creatives go this way in the future.

AI is not only taking, it is also giving 🤖

Here is Zeng’s virtual collection that sparked it all

Useful AI tools for your business or work

  • Whisper Memo is an app for your iPhone that lets you record memos on your phone, and the AI automatically transcribes them and sends it to your email. A super handy app to have on your phone. Now you can record your bright ideas when they appear and get them written and sent to your email by the time you are back from the store. I use mine all the time! So easy yet so powerful! You can try it for free here.

  • Lexica is an easy-to-use generative AI tool and search engine for AI-created art (used to create the handbags above). Try Lexica here.

  • Slides AI is a tool that will help you create your presentation slides in seconds using AI. You can try it for free here.

  • MakeLogo AI is a great tool that lets you create a logo using AI (not free). Need a quick logo for a test project? Get one here.

AI experiment of the week

This week I am stealing someone else’s mojo.

I am sorry to the original creator. I do not remember where I found this screenshot, so I cannot tag your creativity.

This one made me laugh…

The latest robotics finds

The robots are coming… (It was just a matter of time).

Quick bites - interesting AI news

  • Amazon enters the AI space. HuggingFace and Amazon Web Services (AWS) enter into a long-term strategic partnership to make AI open and accessible to all. I was predicting that Amazon would join the battle, and I was right.

  • A human has beaten an AI in the world's most complex board game, Go. However, to win over the AI, the player got strategic help from another AI that had previously played and analyzed the AI's training and found its weaknesses. It then came up with a plan a human player could execute. Is there any hope for us…?

  • The media is full of coverage about how one of the largest sci-fi and fantasy magazines, Clarkesworld, has stopped accepting new work as they are flooded with submissions generated by artificial intelligence. A glimpse into a dystopian future? You can read Clarkesworld here.

  • Research company Baird has created a list of 50 companies they believe will benefit from the AI “tidal wave.” Here are some of the companies mentioned: Workday, Pinterest, Ceridian, Paycom Software, and Nvidia (which I covered in a previous newsletter). Read the full story over at Yahoo.

An ask

I really want to get this newsletter out to more people.

I think that knowledge about the commercial use of AI beyond ChatGPT is essential for businesses and humans in general.

I would greatly appreciate it if you would share this newsletter with someone.

Forward it, share it on Twitter, Facebook, LinkedIn, or whichever is your preferred social media.

Good robot karma awaits you 🤖

A few issues back, I asked if I should switch to a twice-weekly schedule. Although there was about a 60/40 for “yes,” I have decided to stick with the format.

My focus is research and “deep dives,” so I decided to keep things as they are. I already spend about 20+ hours writing this one, and I have another newsletter brewing in the backroom… (not about AI 🙂)

If you want more AI news in your inbox, these newsletters focus on shorter or curated content: Prompt Pulse, The Status Code, & Ben’s Bites.

If you have any questions about AI or any feedback, just hit reply or tweet me @thomassorheim 

That’s all for this week.

Have a great Thursday!

I am off to Spain in an hour, but AI’ll be back next week.

Why don’t marketers like trampolines?




They’re scared of high bounce rates.

This is not the end. It is where the fun begins!


or to participate.