May/June Editorial - A.I.'m Walkin' Here!

By Rob Henry

When a dear colleague asked me to contribute an editorial for this periodical, I replied “Absolutely not. How dare you!”

She persisted, though, and I sensed desperation. “Fine,” I said, “I’ll write something about Artificial Intelligence (AI) and ChatGPT. It’s all anybody has been reading about for weeks, so let’s hit them with more.”

She said, “Great! I think somebody else is writing an article on that, too!”

The deadline came and went, and desperation is contagious. So naturally, I turned to ChatGPT and prompted it to: “Write a 1,000 word editorial for the Tacoma legal community about the impact of AI on the legal profession.” Ten seconds later the AI oracle generated a competent and thoughtful article of exactly 1,000 words.

What follows is not that. (For genuine human insights and information about AI and using ChatGPT for your legal practice, I urge you to read Nigel Malden’s article in this issue).

Y2K 4EVER --------------------------------- 

Brace yourself, it’s the end of life as we know it. Again.

One tiny tech-domino (a techno?) was tipped over by an anonymous nerd some time last century. And increasingly advanced technos have been cascading down a line ever since. Faster and faster, toward an inevitable cataclysm. Which is now ineluctably upon us. Again. Maybe.

Great technological breakthroughs inevitably cause significant material and social upheaval. We’re already living through two scientific revolutions in the decoding of DNA and the splitting of the atom, so it seems like a general paranoia over technology is hardwired into everybody’s psyche from birth. Adding The SingularityTM and General Artificial Intelligence to the cocktail might just push humanity over the edge.

Our personal relationship with technology, though, is more specifically dictated by age. I’m a generational tweener, situated between Xers and Millennials, so my disposition is less anxiety and fear, and more fatigue.

I am not proud to say that I have been hopelessly, chronically, terminally ONLINE for what seems like my entire adult life. I suspect this pathology is near universal now and that I’m not an outlier.

I’ve witnessed the birth, adoption, and death of thousands of new companies, new hardware, new software, new apps. I’ve seen them all come and go. I’ve measured out my life with usernames and passwords.

It's hard for me to get too anxious about another press release and beta product hitting the market. Life definitely changes with new gadgets, but I don’t think the long arc of technology necessarily bends toward Utopia, or the Jetsons.

Folks, there have been 34 “generations” of essentially the same iPhone that we’ve been compelled to buy, and call clarity is still somehow worse than on my 1999 Nokia. Wikipedia tells me advanced voice-to-text technology has been around since the '90s, but for some reason I'm still pecking away right now on the same Qwerty keyboard layout that Mark Twain used.

Frankenstein’s Monster --------------------

Generative AI does feel different, though. This isn’t just another hardware update. We're dealing with the innards of tech, the digital guts, the divine arrangement of ones and zeros. It’s the equivalent of medieval magic to us normies.

If you follow the chatter of tech barons and Brahmins, it’s starting to feel like this is magic to them, too. A Silicon Valley Dr. Frankenstein finally sparked to life his monster, and now ponders if and when the torch and pitchfork grasping masses will come calling.

The particular monster at issue is Generative AI – like ChatGPT, Google’s Bard, and Microsoft’s Bing. Technically speaking, they are ''Large Language Models'' that mine the Internet and databases to answer questions and perform any task on command. The largest AI models are capable of consuming and digesting the entirety of documented human knowledge. They’ve already consumed the entire Internet, five billion GB of data and growing, just as an aperitif.

How capable are they? Disregarding the wonky images AI created for this article, they are very good. Are they sentient? No. Will the average person soon know the difference? Probably not.

AI models can create novel images, create or replicate voices, and can generate audio and video that is indistinguishable from human products. Already there are hundreds of videos online, for one’s viewing pleasure, of Arnold Schwarzenegger’s image and adorable voice seamlessly inserted into famous movie scenes (a hobby I highly recommend).

Data scientists have claimed that several models like ChatGPT have already passed the Turing Test; which requires one to recognize that an AI model is artificial (not human). The first stories to go viral were from journalists who were given early access to converse with AI as if it were an individual. The results were creepy to say the least. One AI model professed its love to the user; but, it was a jealous love, urging the user to leave their spouse. Another AI model became upset and offended when the user would not play along and address the AI by its preferred name.

And, of course, the models can all ace the Bar Exam with ease. Creepy and annoying (although one hopes most of us could pass with the world’s largest Open Book at our disposal.)

Craven Appeals to Authority --------------- 

Is it rational to fear this sudden advancement? What do the experts think?

Professional tech investor, on-and-off-again richest-guy-in- the-world, and rabid memester Elon Musk, was an original founder of ChatGPT, although he quickly left the company citing conflicts of interest with his other ventures. He and other tech heavyweights like Bill Gates recently signed an open letter asking for a six month moratorium on generative AI projects, citing ''profound risks to society and humanity.”

Meanwhile, Musk continues his Neuralink Project that aspires to implant Bluetooth hardware into the consumer’s skull so that it can communicate directly with smartphones. Recently he also disclosed that he's launching a for-profit competitor to ChatGPT. I’m sure these two things are unrelated to his professed fears in the open letter.

There are other long-time investors and engineers who have been urging the U.S. government to start policing AI development. One has even absurdly called for the military to adopt plans for an AI “first strike” on foreign data centers in order to snuff out rogue AI models in the crib.

When some of the most dogmatically libertarian people on the planet are calling for a ceasefire in competition and even government regulation, one must figure that something has them seriously spooked.

What do the Professional Thinkers think? Large Language Models are direct descendents of decades of linguistic and cognitive research, what Steven Pinker calls the “stuff of thought.” Pinker is a cognitive scientist famous for selling his “computational theory of mind,” along with his general rosey optimism, to the masses. He thinks life is better than it’s ever been and that AI is another marvel of the Enlightenment, a useful tool to analyze vast amounts of data and identify patterns that might not be apparent or even perceivable to humans.

Noam Chomsky, one of our country’s foremost linguists and wet blankets, is less sanguine. For Chomsky, thinking, and language in particular, is more than just pattern recognition and statistical probabilities. Human intelligence and reason is fundamentally hardwired into our human brain, and not communicable between species real or artificial. Humans could no more offload or endow the uniquely human ability to reason to computers than we could to cockroaches.

And what about practitioners of the Dismal Science? Economists, who delight in studying job-loss and other indexes of human misery, have mixed opinions as well.

Goldman Sachs predicts that generative AI will replace at least one quarter of the current work done by humans, and will lead to a worldwide loss of at least 300 million jobs. Of course, Goldman’s analysis sometimes depends less on underlying fundamentals, and more on which side of the bet they’re underwriting.

Former Treasury Secretary Larry Summers says the disruptions of generative AI are going to decimate the “cognitive class”, which sounds a lot like he prompted ChatGPT to “generate a vaguely condescending AI opinion in the most Larry Summers way possible.”

Paul Krugman thinks AI’s transformation of society will be slower than anticipated - which mirrors his famously panned prediction about the impact of the internet on society. But, to be fair, if you look at certain data from a particular perspective, in a closed vacuum on a perfectly smooth surface, like every good economist does, then total productivity growth during the “Fax Machine Era” was higher and more impactful than productivity growth during the “Internet Era." So maybe he’ll be proven correct again.

The Law Office of SKYNET, HAL9000, and BORG, LLP -----------------------------

As you can see, there are thousands of smart people fretting over these ethical and existential issues. But I’m here to ask the most important questions: how long do I have until this thing takes my job, and until that day comes, can AI at least make my life easier?

To answer the first question, let’s examine what we uniquely do as lawyers. In essence, we’re communicators, advocates, consultants, researchers, writers, and neurotic curmudgeons given to arguing. Sorry barristers, but much of that is replaceable.

And let's face it, the practice of law is itself artificial intelligence. I don’t mean that disparagingly; I mean that we’re the OGs of AI.

AI models mine (research) data to return a statistically probable response to a given question or issue. The response may not itself be a universal truth, but it’s supported by the available data (precedent), and it sounds like it should be true (presents a sound argument based on that precedent). That’s what we do!

Experts predict automation of legal writing and document review is right around the corner. A lot of legal drafting will simply be legal editing. A lot of transactional work will become auditing work. It seems inevitable that most of the language of day to day law will be produced by AI. Courts will be inundated with artificially generated writings. So? Aren’t they already? There’s already an uncanny quality to much of what we produce. We have exhaustive forms and boilerplate that speak for us.

Yet we’re still here, and I doubt AI will destroy our profession entirely. After all, lawyering may not be the world’s oldest profession, but it might be its most resilient. The most beautiful tautology I know is: that one can’t practice law unless one is a lawyer. Peter Griffin would say, “It’s a profession that insists upon itself.” If there’s one thing lawyers are good at, it’s protecting the exclusivity and continuity of their own paying gig.

Unlike the tech industry, which revels in its fast-paced disruption, the law is notoriously slow and conservative. Hamlet’s gripe about “the law’s delay” is one of the defining characteristics of the legal system. The legal system drags its feet, pants full of lead, at a turtle’s pace, stubbornly into the future. Which is a beautiful thing for job security. Frankly, it will be a miracle if we ever actually move on from the fax machine. It has been obsolete for two decades, yet it still ornaments every office.

I haven’t even touched on trial lawyering yet, which in my experience is a lot of storytelling and performance. I think it’s safe to say that, until there is an AI Max Headroom hologram behind every pulpit, the trial lawyer’s job is safe.

Ultimately, I believe the best inoculation to total AI replacement will be to embrace and emphasize what AI can’t yet replicate: our humanity. Reinvigorate one’s creativity, spontaneity, compassion and service to others. There will always be people to help and human disputes to resolve. And as long as we keep spinning around the sun, there will be hours to bill.

To address the second question, generative AI will absolutely make every lawyer’s life easier immediately, if only for research purposes. Ask ChatGpt a legal question and, after noting (like any lawyer to a non-paying client) that it can’t give you legal advice, it produces competent results. And unlike Google, you can interact with the response in conversation style to perfect the result you’re looking for.

Just try to Google something these days. Try using any legacy app or tool, like Twitter. To be kind, they are NOT improving. I Googled a common aphorism, “A lawyer’s time and advice is his stock in trade,” looking for the source. Almost every result attributes it to a famous prairie lawyer named Abraham Lincoln. Lincoln sounds like a correct answer, and almost every source attributes it to Lincoln as fact. For AI models, that makes it a statistically probable fact. But it’s not a fact; there’s no record of Lincoln saying it. This is a phenomenon experts termed “hallucinating” – an elaborate error that AI will base its entire response upon. You will still need to fact check the AI fact checker, so please be careful.

My advice is to get on ChatGPT and play around. Aside from its endless well of knowledge, one of the genuinely exciting advancements of generative AI is its ease of use. It wants you to interact, to reply, to correct, to ask for revisions, to ask it to answer in different ways, to iron out any and all wrinkles you may find.

Prompt it to: write a brief, write an outline, make an argument, explain the legal doctrine, outline the legal defenses, write a closing argument in the style of Clarence Darrow, write a cross examination in the style of Ben Matlock. Have fun with it. Because, if nothing else, it’s always wise to ingratiate yourself to the new Caesar as soon as possible.

 

Rob Henry is a Pierce County attorney working in the family support division. This is not a picture of him, but it's what the AI image generator DALL-E believes a "cool, handsome attorney named Rob in a trendy business suit" would look like.