In a meeting about AI & design this morning, a clever, senior colleague opened a discussion about whether we could or should use “Bob” as a verb (we work at IBM so our AI IDE is IBM Bob). Even though the original discussion was about gendering and anthropomorphizing software, an interesting nuance, the discussion led to trying to define what “Bobbing” even means, and does it have a positive connotation? What does it mean to “Bob” something?

In the last 6 months, everyone I work with has started using IBM Bob to get their work done. It’s a fantastic tool for coding, design, content, and most knowledge work. Of course, we also can’t ignore the fact that everyone is being encouraged to use AI in everything. As we all adopt this new generative AI tool, what does it actually do for us and to our work?

The Volume Problem

The default first thing we tackle is more stuff, faster: I did <x> assignment 3.5x faster, which enabled me to do 2x more work. That kind of thing is great for performance reviews. Every person on a team can put this kind of productivity gain down as a positive. Increasingly, however, there is a lot of pushback against the “gains” we’re all supposedly getting from AI.

If the only real change is volume, did everything actually get worse? If a company has 30 products, each with their own docs, marketing content, and product pages and everyone starts putting out more content (more stuff that can’t be maintained, can’t easily be searched, and immediately reads as six-fingered writing) is it actually negative productivity? Even if it’s good, is more an inherent problem? Is anyone better when their days are full of countless well-intentioned proposals, executive summaries, and insightful pieces? There has to be a limit to what we can all take in, and the internet is starting to feel more and more like something not intended for humans. If both the speed and volume of content, even good content, are so much that no one can take it in, then what was gained?

But Is It Any Good?

Also, is all this content any good? When a colleague sends me an impressive-looking site, repo, or slide deck now, the natural question I have to ask is, “Is this actually good, or is it the result of 30 minutes of work and 13 Bobcoins?” AI is just so good at churning out anything at a mind-boggling scale, how can we tell when someone put effort, testing, and reflection into an idea? Everyone’s Slack feed is now full of “product announcements” and it’s difficult to judge the pressure-tested ideas versus something that was just…generated. As an example, the posted version of this piece will be roughly the sixth I worked on with Claude, with Claude pointing out I’d overused “really”, miscounted some ideas, had stray commas, and a few missing words. More importantly, Claude rhetorically called out weak arguments, some I still haven’t resolved. For example, even if this post is decent, who really cares? Is the world better because Claude can now sit down and guide just about anyone into writing engaging content? Also, what someone actually knows about a thing can’t easily be verified, something educators and professors are wrestling with.

Does the Creator Understand It?

It’s currently all the rage to find slop out in the world: LinkedIn posts, blogs, commercials, slide decks, photos, videos. I dislike a soulless internet as much as the next person, but for me, the challenge isn’t “did an AI make this?”, it’s “does the creator understand what was made?” If someone reads a book, listens to 5 podcasts, does some research, watches a few videos and then uses AI to collect their thoughts on it all, I’m a fan. I don’t really care if their words are entirely their own - it matters to me that they could expound on all their thinking.

On the other hand, if someone feels they’re keeping things raw and punk rock by not reviewing their work and posting a bunch of random thoughts and proudly displaying grammatical errors, that doesn’t feel punk rock to me - it just feels lazy. At the same time, if you spent 20 minutes over the weekend coming up with an idea for an app that isn’t tested, can’t scale, couldn’t be maintained, then maybe it’s not time to share it. I hope we don’t end up in a world where we have to choose between off the cuff, unreviewed “human” products and perfectly polished, but banal, artificial slop.

How I’m Going to Write

I’ll be using this sounding board more consistently now that I’m starting an MS in AI in Business at Boston University, so I’m trying to formalize how I’ll write. Naturally, I’m trying to navigate my way through the world of artificial slop, so here is the workflow I plan to use: Every first draft is my own, but I’ll be using Claude to refine my thinking. Hopefully, this will make my voice authentic but refined. I’ll test over time and see how it goes.

Where I’m Standing

I also recognize that I have a privileged position when it comes to all this: I read books in a mostly postliterate society. I sort of know what I’m doing. I taught high school English for the better part of two decades before transitioning to technical content, a field that is being turned upside down on an almost weekly basis. Lastly, my native language (in fact, the only language I speak) is popular. I fully respect that there are tons of folks out there with a lot of technical knowledge and that language is the barrier to them being understood.

No Natural Intelligence Owns It

The cynical read of the verb “Bob” is probably something like “doing something I don’t really understand.” In other words, here is a process I don’t comprehend, a passage I haven’t actually read, an application I can’t scale or maintain, a design I can’t iterate on. If that’s the read, then it’s natural to feel revolted when we encounter AI-generated anything: it’s soulless not necessarily because an artificial intelligence made it, but because no natural intelligence owns it.