Building Trust in AI: From Skepticism to Scaled Success
As generative AI matures, expectations are evolving. The question is no longer “Can we use AI?”—it’s “How do we trust it enough to scale it?” Many organisations are moving beyond…
New York NY — August 23, 2017 — ILTACON 2017 is now over. The tent has come down, the acrobats are back on the ground, the clowns have gone home. Las Vegas may be a circus, and we did have a few parties, but ILTACON is a superbly substantive event—more than 250 educational sessions, 350 speakers, 1800 members and 2000 business partners attending. And truly peer-powered by the Conference Committee and scores of volunteers, with deft and indefatigable assistance from ILTA’s staff.
Measured by the energy and attention in sessions and in the hallways, ILTA evidently agrees that “AI is the new electricity,” in the words of Andrew Ng—whose c.v. includes stops as founder of deeplearning.ai and Coursera, head of AI research at Google and Baidu, and professor at Stanford.
AI is not pizza—you can’t just order it up for quick delivery (“30 minutes hot or it’s free”). As Adrian White, the CIO of Wilmer Hale put it, when partners call to say “Get me some of that AI,” we need to say, “Let’s talk about the problem to be solved, then we’ll talk about solutions.” This year, the crop of real and useful solutions to important problems was better than ever.
Some impressive examples.
Yes, from a technical perspective the “just software” point is true. The ever-more complex algorithms in AI are indeed software and they are written by programmers, though lately some algorithms have been writing other algorithms, a step toward the self-replicating machine. And this sort of de-hyping is useful: I suspect we haven’t reached the peak of the Gartner hype slope.
Yet from a practical, prudential, business perspective, as AI moves closer to the center line of what lawyers do, the dismissive “it’s just software” line is dangerous. We need to know what we’re talking about, and what our algorithms are doing—for us and our clients.
It is probably true, as Frank Chen of Andreessen Horowitz said recently, that “in a small handful of years, software without AI will be unthinkable,” just as for decades software without SQL behind it was (almost) unthinkable. Nonetheless, it is also true that “no rational law firm is going to outsource thinking,” as one law firm CIO said. Thus the AI challenges of transparency, interpretability, and auditability are particularly important to lawyers.
What else did we hear and learn about AI?
It is important in all legal technology to keep an eye on the practical amidst the pretty—whether the domain is mobile device management, security, or ediscovery. That is doubly so in AI.
No doubt there were others talked about or demonstrated that we did not get to. This list excludes the excellent AI work being done for electronic discovery and legal research, as well as IBM’s Outside Counsel Insights (OCI) and other “business of law” work.
In short, 2017 is an outstanding year to see AI doing useful work in the law. And it’s only August.
Next year at ILTACON 2018? We forecast less talk about “what is AI?” and more working AI examples built upon well-defined statements of problems and solutions. Here at Neota Logic, we are working with clients on many new applications and at the same time are charging ahead with improvements and extensions to our AI-based intelligent automation platform for expertise, processes, and documents. More to say this autumn.
Yes, ILTACON 2017 did sometimes feel like a circus—the allée of themed slot machines from Wonder Woman to Willy Wonka (for children?) set the tone. Yet circuses can be illuminating and inspiring as well as entertaining. And fortunately, there was Yoga every morning at 6:30, or Zumba for the fast movers.
Missed ILTACON? See key takeaways from Neota Logic’s @michalemillsny
