IQ points of models are not the current blocker. The current blocker is actually answering questions like: How do we take this intelligence and apply it efficiently to a product, to a workflow? Are we actually able to do the things that the end users want us to do?
The promise of AI is often centered on its ability to perform complex tasks faster and more efficiently than humans. AI’s abilities may be stretched every day in new directions, but that doesn't always mean its newest talent will actually serve users. It begs the question: are we building AI to solve real-world problems, or are we simply adding new capabilities because we can?
Adrian Parlow, Co-Founder and CEO of PointOne, a company that has developed an AI-powered solution for timekeeping and billable hours in law firms, argues that many AI solutions miss the mark by focusing too much on LLMs' newest talent versus leveraging it to address pain points for users.
Eye on the prize: "I think a lot of the AI solutions being built are thinking the wrong way. They're thinking, 'What can LLMs do? Let me build that' versus ‘What do users want? Let me figure out how to build that with LLMs'," Parlow explains. He believes the real opportunity for AI isn’t in replacing human expertise, but in automating the tedious, low-value tasks that people hate doing.
"It helps that we're trying to automate the thing that is the number one most hated task of pretty much every attorney," he says, in reference to timekeeping—something that lawyers dread but is essential to their practice. PointOne has successfully addressed this pain point, creating an AI solution that not only makes life easier for lawyers but also boosts efficiency for law firms.
We’re not going to have AI judges making decisions, but the judge might ask the AI to evaluate something or for some context.
The real challenge: Parlow emphasizes that the intelligence of AI becomes irrelevant if it isn't applied well to daily workflows. "IQ points of models are not the current blocker. The current blocker is actually answering questions like: How do we take this intelligence and apply it efficiently to a product, to a workflow? Are we actually able to do the things that the end users want us to do?"
AI shouldn't replace talent: This user-centric approach is evident in PointOne's solutions, which go beyond timekeeping to tackle other back-office inefficiencies in law firms. Parlow believes that rather than pushing users in a direction they don't want to go, companies should ask, "What direction do they want to move in?" By focusing on areas that users wish they could outsource versus replacing their work in areas that demonstrate their talent, products are much likelier to have higher usage rates and stickiness with customers.
Ethical balance: Looking forward, Parlow acknowledges the ethical concerns surrounding AI, especially when it comes to replacing human decision-making. However, he envisions a future where AI acts as a "smart sidekick" to professionals, providing them with insights and ideas but never fully replacing them. "We’re not going to have AI judges making decisions, but the judge might ask the AI to evaluate something or for some context," he says. "Similarly, we're not going to be having AIs negotiate against each other on mergers, but you can be pretty well assured that the partners doing those mergers are going to be asking their AI sidekicks what the market looks like. It's just going to level up."
