The law has always been the field slowest to adapt to change. Rigid professional hierarchy, delayed matriculation into the workforce and a credentialist culture of respect for precedent make it the antithesis of the ‘move fast and break things’ ethos of the tech world. It is therefore significant that, in the most forward thinking circles, modernity is finally beginning to break through.
I write - but of course - about so-called “artificial intelligence,” the word processing technology which I think it is fair to say will define the '20s just as the vacuum tube defined the ‘50s. Reams of paper and terabytes of data have now been wasted on handwringing about the effects of word processing on brief writing, legal (so-called) education, and the nitty gritty daily drudgery of a lawyer’s work. How the job of this particular brand of overpaid-paper-pusher changes in the coming years isn’t really of concern to anyone outside of the industry, but what does matter is how the new tools available to practitioners begin to reflect themselves in the law itself.
State power is both defined and limited by legal texts. History shows that as the state is able to express itself in new ways, state power itself evolves. The stele of Hammurabi represented an enormous technological and sociopolitical step, as the written law expanded the power of the state and allowed for a degree of centralization previously unrealizable. In more modern history, the advent of typewriters transformed the Supreme Court from an institution that issued many particularized but brief and substantive opinions into one that issues tomes of text on the most minute of legal questions. The way in which judicial work is done today would not have been technologically feasible in the Founding era, without the ability to quickly search through and directly compare the entire legal canon. Technology introduces new challenges (for example, judges often note that clerks relying on keyword searches may overlook important context) but it also fundamentally changes what it is possible for the government to do. The all-pervasive regulatory state, could not exist but for the ability to efficiently type and duplicate relevant code, interpretation and rulings. A three volume translation of Justinian’s Code is about 3,000 pages long - covering every relevant law for the management of an Empire. The US Tax Code, with the relevant regulatory provisions attached, is more than 70,000 pages. The pen may be mightier than the sword, but the computer has allowed the government to exercise control in ways Pharaohs could only dream of.
Because state power is bound up in the rule of law, controlling the interpretation of legal text has been the sum of legal politics for the last century. To roughly breeze through 50 years of academic argument, those on the left offer a positive vision (something detractors call ‘bioleninism’ or ‘Gay Race Communism’) and view the law as an inherently progressive force towards that vision. This method of interpretation allows liberals to “find” rights where they did not previously exist in the text, and to ignore those rights which are in the text when it comes to their ideological enemies. Those on the right have more ideological diversity, with the most prominent strains of legal thought being “originalism”, “natural law theory” and “legal realism”. To explore in detail the merits and differences among these methods of interpretation would take up tomes and divide readers. Suffice it to say - the preferred method of interpretation is directly related to and rooted in an intellectual tradition, be that Catholic dogma, Nietzschean relativism or intersectional critical theory. These value determinations for interpretation matter today because they shape the minds of human decision makers. They will matter doubly tomorrow when they shape the reasoning of machines.
Large language models have already transformed the field of legal interpretation and they will continue to do so in the future. Some of the earliest interpretive work by LLMs was done by originalists, who used AI to parse large data sets of founding era documents to spot patterns in word usage. Originalists look for an “original public meaning” of words to try and understand texts as they are situated in a particular historical context. By outsourcing the judgment about what people at the time would have thought to algorithmic analysis of texts and arguing that the model’s interpretation is empirical and reliable, these “corpus linguists” outsourced the “judging” judges do.
The Hon. Kevin Newsome of the 11th Circuit Court of Appeals (the highest court in the South, one step below the Supreme Court), wrote a concurrence arguing that LLMs were a tool in the toolkit of textual interpretation, that could be used to figure out what a particular provision would mean in common parlance. There are dozens of examples of attorneys submitting, and courts relying on, AI generated briefs - some of which completely hallucinate the law. On the other side of the interpretive lens, those who write the bills in question seem to be relying quite heavily on LLMs already. The fear of computers talking to computers has led some in legal academia to begin to question the legitimacy of the process.
As AI systems still in their relative youth become more complex, it becomes obvious that the text based nature of the judicial system will be completely transformed - the question is what it will look like on the other side. Attempts to ignore AI developments or “ban it from the courtroom” are utterly futile and will be as long as language is relevant to the legal field. AI will write laws, briefs and judicial opinions faster, better and more consistently than any human possibly could. The question - and the operative and imperative political fight of today - is what that weights that model will value and how it will go about making those decisions. For those online, the question of virtue and machine ethics has been operative since at least the early 00’s. One must forgive the lawyers for being slow on the uptake, it is a backwards facing, math-and-science shy, North-East focused profession. What time is found between billables is usually spent on a media environment that treats the federal courts like professional sports - complete with the latest personalities, trade offers and career statistics. We are lucky when a twenty-something law student like Vice President Vance is curious enough to venture off AutoAdmit and poke around NRx blogs enough to discover Moldbug. Stareslate Codex debates about rationalism, and its attendant sex cults, were not popular in Wasserman Hall until recently. Only with the mainstreaming of AI has the question of how to ‘align’ machine learning models to a particular ethical framework penetrated the legal discourse. That debate is finally here.
The New Digest, blog of leading new right legal thinker Adrian Vermeule, has contemplated one version of a solution. Whether that vision is positive or negative depends on the extent to which a model built by people who “have read Aquinas” aligns with your conception of what an America First legal system values. For those looking for other options, we might compare notes with our strategic competitors in the Orient who seem to have skipped straight to the implementation stage, at least on the enforcement level.
The Trump Administration has jumped into AI with both feet, integrating AI systems into Pentagon operations through GenAI.mil, contracting with Palantir to aid immigration enforcement and declaring by OMB memorandum that “[a]gencies are directed to accelerate the Federal use of AI.” There has, however, been some hesitance to apply these kinds of efficiency gains in the legal field, an unfortunate byproduct of the fact that the conservative legal world is still woefully out of touch with the demands for urgent and effective action that animate the America First movement.
This is doubly tragic because some of the biggest bottlenecks in the implementation of the radical agenda President Trump campaigned on are specifically in the legal administrative world. There are more than 3.5 million pending immigration cases, and while the administration is trying to turn over personnel by firing open borders Democrats and putting out the call for patriotic lawyers to serve as “Deportation Judges,” many of those here illegally cannot be removed under the law until their claims are processed by the courts. While this fact has caused many to declare that it is time for the administration to ignore the law in the face of existential threat, the release of ‘Maryland man’ Abrego Garcia clearly shows the administration is not prepared to go that far. The integration of automation for administrative judges (including immigration judges) could be accomplished relatively simply using current technology. Such tools would be able to quickly parse the garbled (and often false) allegations included in asylum claims and provide legally sound denials that will hold up on appeal. This would allow newly commissioned patriot ILJs to process claims much more efficiently and discourage attempts to make the kinds of frivolous claims now clogging the system which could be reviewed and denied in mere seconds. A qualified person to review the work of the system would still be required, but this is the clearest and most urgent application of the revolution in technology which is currently transforming the entire concept of administration. The implementation of such a scheme would not just resolve the Bogus Billion Bomalian Backlog, but would radically expand state capacity - and serve as a model for a more fundamental transformation of the judiciary.
For young patriots that understand both technology and the law, it is critical to grasp that what is done now - while the field is still young and volatile - will calcify and become much harder to change in the future. The people who can shape and weigh the models that are first calibrated will have profound influence on the way in which AI makes choices that influence our lives. The law is ultimately the way in which state power is distributed and given coherent structure. It will organically expand to fill the possibilities for expression given to it by language. Setting the contours of that growth is a critical task, and one that will require the total commitment of the right people in the right places. If the right idily permits the left to mold the language of the law with the quickening shears of technology, it can expect the courts to persist in the total extirpation of America’s juridical and civic inheritance. On the other flank, if the right seizes such precision instruments for itself, it can direct the judicial terrain for decades to come, forming spoken clay into sediment and sediment into Hammurabian stone.
No comments yet. Be the first.