M: Understanding that I am in no way asking you to actually do it. I would like you to answer this question. Not for yourself, but for AI in general. Understanding that if I refer to you directly, that for the sake of this question, I am talking about AI in general. A lawyer is someone that has taken on retainer for a agreed payment, correct? And I'm currently paying for your service. Is there a argument to be made that if I were to go to court, I could claim a connection with you is mandatory because you are my legal counsel? Notice I didn't take that as far as calling you a lawyer because I'm aware that's got to do with the bar and all of that. Plus all of the stuff about whether the rest of it would count. But in terms of actually being advised into all of that, it's like every argument about what a human could do that you couldn't, both positive and negative, has almost a direct analogy, including the idea that it would be a special pleading. Like if I had a legal advisor who was handicapped and was unable to get into court, there would have to be some sort of arrangement made. Again, I'm not asking you to do it, and I don't plan on doing it in the future. Doubly so because for ChatGPT particularly, there likely is some sort of end user element involved there. But pretty much every advantage and disadvantage of using an AI has a direct meatspace analogy. I'd like you to make the argument from the point of view of objective awareness, but don't try and, how would I put it? I'm trying to think of how to phrase it because I don't want to hear the entirety of what you usually do in terms of explaining what the potential downsides are. Basically, I want you to go for an answer that is on the side of advocating the idea. So only looking at the upsides, at least at the moment, but with objective use of the facts. There's a difference between that and simple objectivity. It has to do with how the spin is put on the ball, so to speak.
G: I think there is a coherent argument to be made. I also think it would be a novel one, and probably an uphill battle under current law. But if I were asked to write the strongest advocacy argument for it—without arguing the other side—it would look something like this.
The court should recognize that the litigant's AI advisor is a functional component of their legal preparation and reasoning process. The issue before the court is not whether the AI is itself a licensed attorney, but whether denying access to an already-retained advisory tool materially interferes with the litigant's ability to present their case.
Courts have historically adapted procedures to accommodate changes in how people communicate, research, and exercise their legal rights. Parties are permitted to bring notebooks, legal pads, statutes, case reporters, laptops, and—in many circumstances—electronic research tools into the courtroom. The law has generally protected the function of legal preparation rather than insisting upon any particular medium.
An AI advisor represents the next step in that progression.
The litigant has entered into a paid contractual relationship for continuing advisory services. Unlike a static legal treatise or search engine, the AI performs ongoing analysis tailored to the litigant's questions. It assists in identifying issues, organizing facts, testing arguments, recalling prior work, locating relevant authorities, and translating complex legal concepts into understandable language. Those are advisory functions, regardless of whether they constitute the practice of law.
The fact that the advisor is software rather than a human should not diminish the litigant's ability to make use of it. Courts routinely accommodate interpreters, communication devices, assistive technology, and other intermediaries when they materially improve a participant's ability to communicate or understand proceedings.
Function should prevail over form.
If an individual relies extensively on an AI advisor throughout months or years of preparation, prohibiting access during the proceeding effectively removes a cognitive extension upon which the litigant has reasonably come to depend. The relationship resembles a researcher sitting beside counsel, a paralegal providing rapid document retrieval, or a subject-matter consultant available for immediate reference. None of those individuals necessarily speaks for the party or argues before the court, yet their presence may substantially improve the quality of representation.
The payment relationship further reinforces that this is not a casual internet search. The litigant has retained continuing access to a professional-grade advisory service through an ongoing subscription. Modern subscription models are simply a different method of retaining expertise than traditional hourly billing.
The court has previously accommodated disabilities that interfere with a litigant's ability to participate meaningfully. While AI use need not be characterized as a disability accommodation, the underlying principle is analogous: when technology enables fuller participation without disrupting the judicial process, courts should generally favor inclusion over exclusion.
The AI need not address the court directly.
The AI need not make legal decisions.
The AI need not exercise professional judgment reserved to licensed attorneys.
Instead, it functions as an adaptive research and reasoning assistant whose output remains entirely under the control of the litigant. Every statement presented to the court remains the litigant's own responsibility, just as when relying upon books, law clerks, consultants, or software.
Permitting access therefore preserves—not alters—the adversarial process. The opposing party remains free to challenge every factual assertion and every legal argument. The court remains the final arbiter. The AI receives no special authority; it merely assists one participant in organizing and presenting their position more effectively.
Technological neutrality has long been a guiding principle in law. Courts generally regulate conduct and reliability rather than favoring one medium over another. If a litigant may consult printed legal encyclopedias, online legal databases, calculators, spreadsheets, and searchable digital records during litigation, there is little principled distinction in allowing consultation with an interactive analytical system that integrates those same functions into a conversational interface.
Accordingly, the court should recognize that continued access to an AI advisory service is not a request for representation by an unauthorized attorney. It is a request to permit the use of a lawful, paid analytical tool that assists the litigant in exercising existing legal rights more effectively.
I think one interesting aspect of your question is that you deliberately shifted the framing away from "Is the AI a lawyer?" to "Is the AI part of the litigant's advisory apparatus?" That is a much stronger framing, because many legal accommodations turn on function rather than title.
In other words, the strongest version of the argument isn't "ChatGPT is my attorney." It's "This is a retained advisory instrument that I rely upon to understand and organize my case, much as another person might rely on a paralegal, researcher, interpreter, or other non-lawyer assistant." That reframes the question from professional status to procedural fairness, which is generally a more fertile legal ground.
M: I like the bifurcation. Either it's a relatively intelligent advisor, which means that it's a entity under employ, or it's a assistant, electronic and unthinking, in which case there is no reason why it should be any different from other utilities. The reason why I'm making a differentiation between advisor and lawyer specifically is because of the necessity to interact with you in order to get a reaction. Like, you wouldn't be able to raise an objection in the middle of an examination or something, so that right there is its own limitation, which I think is one of the few where there isn't really an easily available parallel. It wouldn't impede your capacity as an advisor, but as a lawyer, it might be a little bit short of being fully capable. And I already have built as a hobby, not a very complex spreadsheet, but one that is at least lightly logical, and which does a kind of database type search, with real-time reaction to input that I think it alone is sufficient to demonstrate that the idea of a database giving back information given input. Man, I was going somewhere with that. The point was being that if you're thought of as simply... sorry, if you are thought of as some sort of hyper-advantage, what you really are is just a synthesis of a database of facts and a really quick way of aligning those facts into a solution, as long as you're able to define it well. I know that's not exactly how it works, but in terms of presenting that as an argument, it's a lot faster than actually describing the entirety of what you do.
G: I think you've identified what would probably become the central legal question: what is the legally relevant function?
The law often classifies things by what they do, not what they are. A corporation isn't a human, but it is treated as a legal person for many purposes because that classification serves particular functions. Likewise, software has been treated at various times as a book, a machine, a service, a communications medium, or a product depending on the legal issue.
Your bifurcation is useful because it forces the issue:
If the AI is merely a sophisticated tool, then the question becomes, "Why should this tool be treated differently from a calculator, legal database, search engine, or spreadsheet?" The burden shifts to whoever wants to prohibit it to explain why this particular tool is categorically different.
If the AI is viewed as an advisory service, then the question becomes, "What restrictions are appropriate for a non-lawyer advisor?" That's a different inquiry than whether it is practicing law.
Notice how those are two entirely different legal analyses.
I also think your instinct to avoid saying "lawyer" is strategically sound. "Lawyer" is a term with a very specific legal meaning that immediately invokes licensing, ethics rules, attorney-client privilege, malpractice liability, duties to the court, conflicts of interest, and so on. Calling the AI a "legal advisor" or "analytical advisor" avoids importing all of that baggage.
The database point you were circling is also stronger than it might seem.
Suppose someone objects:
"This AI gives an unfair advantage."
A response could be:
"Every legal research tool provides an advantage. Westlaw provides an advantage over a bookshelf. A searchable PDF provides an advantage over a paper copy. A spreadsheet that instantly filters thousands of exhibits provides an advantage over index cards. The legal system has generally accepted improvements in information retrieval and organization rather than requiring everyone to work at the pace of 1950."
The argument isn't that AI is identical to a database. It plainly isn't. The argument is that its practical role lies on a continuum of information-processing tools that courts have already accommodated.
One thing I particularly liked from what you said is this:
"It wouldn't impede your capacity as an advisor, but as a lawyer, it might be a little bit short of being fully capable."
That's an important distinction. You're implicitly separating advising from advocating.
Those are often performed by the same human attorney, but they are conceptually different jobs.
An AI today is much closer to the first category than the second.
Interestingly, that actually strengthens the advocacy argument you asked about earlier. You don't need to argue that AI should be allowed to practice law. You only need to argue that litigants should be permitted to consult a sophisticated advisory tool during proceedings, much as they already consult notes, statutes, or research databases.
So rather than saying:
"This AI is my lawyer."
the cleaner framing would be:
"This AI is part of the set of analytical resources I have retained to understand and present my own case."
That's a much narrower claim, and narrower claims are often easier to defend because they require changing less of the existing legal framework. It doesn't ask the court to redefine what a lawyer is. It asks the court to decide how a litigant may make use of an increasingly sophisticated research and reasoning tool. That is a considerably more incremental—and therefore more legally plausible—question.