Discovery, AI and the brain in the jar

July 29, 2023

In the sixth grade, lunch time was a critical hour for survival. It was a time for escape, away from the bullies rounding up young immigrants with pushing, shoving, tripping, spitting and other courtyard tortures. As one of a handful of Central American immigrants at Jonesboro Elementary School in Lee County, I knew that making yourself invisible was a skill that I needed to master.

I slid out of the cafeteria through the emergency exit every day and took shelter in the Dewey Decimal-encrusted library aisles, where even the librarians didn’t realize signs of life could exist, much less grow there. I buried myself in the L. Frank Baum, C.S. Lewis, and J.R.R. Tolkien worlds, then Arthur C. Clarke, Agatha Christie and Ray Bradbury – occupying worlds that were more magical than my current grade school version of “Lord of the Flies.” As I sat cross-legged on the floor, I was surrounded by thick volumes of  Encyclopaedia Britannica, half-opened for searches. Search before search existed on screens. 

Tandy 1000 SX
“Tandy 1000 SX with IBM Monitor” by Ben Franske is licensed under CC BY-SA 4.0.

It was the era of ARPANET and then TCP/IP PCs, and Compaq portable computers, the Tandy 1000, with the MS-DOS black screens against which the librarians typed long strings of white letters and numbers. On the other end were public data networks and other libraries. The librarians had long lists of these sequences printed out on dot matrix printers in neat piles on the reference desk.

I volunteered to type those strings for them, just so I could wait and watch the intermittent blinking square while it reached the other computer across time, space and distance. Perhaps it was this dimension, perhaps it was another, but I would have waited for as long as it took had not the fourth period bell rung for the second time. 

When I think about discovery and the process of finding or learning something for the first time—something unknown or unseen—I think of all those things that require exploration, doggedness in the tracing, and acceptance of our own limitations, of knowing that we do not know. Sometimes, discovery requires leaps of faith into something quite uncomfortable.

The internet changed our experience with discovery, knowledge and our sense of the scale of the world and what was knowable. Sometimes, when I had finished typing the long strings of letters and numbers on the Tandy, in that cold, quiet and dark little school library, I imagined my brain in a jar connecting to other brains in their jars, searching endlessly for the answer to a question and perhaps some insight.

These days, I spend less time in between the stacks of books and more hours in front of screens and recording webinars. Lately, it’s been a lot about artificial intelligence and machine learning. For the last four months, I’ve been experimenting with AI, specifically artificial general intelligence (AGI) and generative AI, such as ChatGPT, Bard, DALL-E2  and Midjourney for images. 

First, I tinkered with ChatGPT personally to help me write letters for friends who asked me for help, but I just didn’t have the time. Then I used it for emails, for example, to a utility company when an electric pole needed replacement and I wanted to strike the right tone and not an angry one. It helped me to better understand a family member’s perspective when we were at odds about a topic or to even plan out my weekly schedule.

I used AI to help me write an outline for this story because I can’t find anything more intimidating than starting with a blank page. I incorporated and tweaked the section header titles and deleted the rest. It was a type of collaborative effort, different in kind from  the use of search engines, that made it possible for me to write this.

I continued to use AI, more professionally this time, to help me edit documents and shorten what I had already written or making things clearer, taking notes, creating agendas, coming up with budgets and emails, and helping me set out a content strategy or engagement plan. I used it to do small tasks that I didn’t want to expend too much energy on. Then I tried something higher order. I started using it to help me figure out a reporting strategy based on a pitch, to help me find a reporter’s nut graph or to find the missing source or perspective in a story. At the same time, AI tools are also flexible enough to help process information faster, analyze difficult topics and help discover in a way that I have been doing manually since I can remember.

“The new change is about the second half of the discovery process. We have refined our search. These new tools help with the second half, which is analysis,” said John Tredennick founder and CEO of Merlin Search Technologies, in a talk called “Five Ways to Use ChatGPT in Investigations and EDiscovery.” He called ChatGPT a brain in a jar because it doesn’t know anything about your particular subject or the documents you’re working with to produce the information that it’s going to process. So you have to connect the documents you’ve collected to this brain in a jar so you can harness the computer and computing power of many computers working in tandem.

For Tredennick, ChatGPT and these large language models are in their infancy, but they are still able to provide incredible power to change the game for discovery, which is not about search capabilities, but the analytical and review side. That is, to analyze, synthesize, and report on—essential for reporters and researchers.

I had already experienced the transformative impact of AI on various aspects of my life, accelerating insights, discoveries, and analysis. It felt like my own secret, like being behind the stacks and entering “The Martian Chronicles” for the first time. Surely, no one had walked and explored this planet before, not as  I had? But it was no secret, in fact, I didn’t fit the early adopter role. But my journey felt singular, involving my own journey of discovery into the use of not just a new set of tools, but also a way to think, to learn and to process the world.

I considered the ethical implications of AI—irresponsible data usage, lack of transparency in algorithms and use, powering weapons of warfare—and the many ways it could do harm by increasing inequality, spreading disinformation and misinformation, and reinforcing biases. I observed and participated in ongoing discussions on how others were using AI in news and other sectors, and I worked to set up ethical best practices. Whatever we do with this now that it’s part of our browsers and here to stay, was the larger question. It was not unlike when the World Wide Web launched in the public domain in 1993 and browsers made the web more accessible.  

This dance of collaboration was familiar. It was one in which we, individually and then collectively, explored the symbiotic relationship between AI systems and human expertise and knowledge. AI is intelligent, but it isn’t conscious yet. It doesn’t have human intuition, creativity or context, all of which guide AI so it can provide analysis. It is “Blade Runner” and “2001: A Space Odyssey,” a future we only imagined when we watched those stories. I am living it and it feels like a new alchemy.

In seventh grade, I learned to make outlines and diagram sentences, dissecting them into visual representations of a sentence’s structure and how words worked. Each word had a respective place, and the words related to each other in a way that I had to draw out. In the process of breaking down and diagramming sentences, I began to understand if there were mistakes in my sentences, which is essential for a person whose second language is English. I wonder what I would have missed if AI had done it for me? What would have been added had it been available for me to use?

This process of learning AI feels like a step-by-step breakdown of lived experience into tasks, ideas, concepts and learning to analyze in a way where I never feel alone or limited in this process of learning. What have I missed? What’s the structure behind these things? 

When I use AI tools to produce something, I tell people because human connection and empathy are built on trust, not computing power to process information or who analyzes best or fastest. It’s not about who belongs here is the one who made it here first. 

AI should not be the new bully in the courtyard that you have to duck out to the library to survive.


This article first appeared on Carolina Public Press and is republished here under a Creative Commons license.

Leave a Reply

Your email address will not be published. Required fields are marked *