Got a free pass at the last moment to the AI Engineer World’s Fair in San Francisco, found a cheap flight and a cheaper Airbnb. So here I am. Snagged a spot on one of the couches in the Expo Hall… am waiting for the crowd to thin out before I head to the buffet line for dinner.

The fact that they made scholarships to the event available based on need convinced me it would be worthwhile to drop everything and go. But this is still very much an industry event. I’ve had the vendor conversations that I need to have. (And was gifted a pair of branded free smartwool socks!)

The nice thing about deciding to go on such short notice is that I’m not here to pitch. More just to observe and listen.

Does anyone really want or need a robot barista? Microsoft appears to think so.

What I am mostly observing is a bad case of tunnel vision. People outside Wall Street and the AI industry itself don’t really understand how much negativity and mistrust exists around these technologies. Activists tend to call for regulation — and while this is needed, it will also put these tools out of reach for independent developers and small startups like my own.

But it hasn’t happened yet.

I am tempted to just put my head down and use Claude Sonnet to build as much high-quality algorithmic generated code as I can, while the getting is good. In all likelihood, that is exactly what I will be doing for the rest of the summer. But I would also like to outline a few ways in which AI can be used to solve problems that are otherwise intractable.

I have identified six:

Accessibility, Community Moderation, Content Curation, Consensus Building, Education, and Employment

(A3C2E, if you need a really clunky acronym.)

• Accessibility. Voice recognition is almost to the point of being usable in place of keyboard and mouse, but not quite. As somebody who lives daily with constant physical pain from repetitive strain injury, the evolution of this technology to allow hands-free coding and web browsing on a range of devices would be night-and-day. It isn’t there yet. But it certainly could be.
• Community Moderation, Content Curation, and Consensus Building. We don’t need more generated artwork, writing, and music. We are drowning in high-quality information already. What we need are ways to make sense of the incredible richness of artistic output already available online, and connect creators with patrons and enthusiasts. Ditto for online communities of all types, and for any type of process that seeks to identify shared opinions and experiences — rather than upvoting the most extreme and polarizing opinions (something that social media algorithms appear to have encouraged, even before the advent of AI). Full disclosure: consensus building through textual analysis is what my recent work centers on.
• Employment and Education. We need AI that creates jobs, and not just for the most highly skilled echelon of engineers. LLM’s are at a really interesting phase in their evolution right now. The best of them are at a level where they can mimic the output of an actual human. That is to say, they can respond in real time to questions and unique inputs, and generating meaningful replies. But they make mistakes. Organizations have a choice: build oversight and review into their AI workflows, or relegate AI’s to low value interactions — users who can’t afford personalized support — and leave them to deal with the consequences. But it doesn’t have to be this way.

Could job descriptions emerge along the lines of “AI Supervisor?” Would it be possible to build marketplaces and add-ons that specialize in matching human editors and graphic artists to put the finishing touches on generated AI content that is “almost but not quite” good enough?Yes, and yes. This is a different set of priorities than the current shibboleth of “AGI / Nobel Prize Level Intelligence” but it could be applied in practice with genuine and immediate results.

As remote learning becomes more commonplace, the most valuable role of AI may not be as a search engine that provides results in complete sentences. Rather than an “encyclopedia-on-demand,” correctly trained and tuned models could function as a source of input, guidance, and feedback for students working independently, particularly adult learners and those re-skilling. Yes, in an ideal world most of us would like a human tutor or mentor to work with us, but for people who cannot afford to take time off or take out many thousands of dollars in loans to pursue job training, one-on-one AI training could become a valuable counterpart to video lectures and online multiple-choice quizzes.

The challenge is that none of the six categories above are easily monetizable. We are talking about meeting needs that are currently going unmet — rather than using AI to slash costs (in other words, eliminate jobs).

Given the timetable and realities of grant-based funding, charities and the public sector are extremely unlikely to lead this charge.

So what is to be done? I would advocate that it’s time for the industry itself to step up and invest in social impact projects.

This is not an exhaustive list. Other people may be able to dream up further use cases that are equally valid. But it will take more than a hack-a-thon or a single afternoon to generate meaningful results. As a coder who hails from a UX background, I am a firm advocate of user research. Don’t build vanity projects. Instead, find partners and stakeholders.

Without visible evidence that AI is contributing something positive to our world, public retribution will be swift and absolute. Regulation won’t end inequity or violations of privacy from AI, although it may curtail the worst abuses. It will lead to something more like a guild economy, and render another sector of software (like cryptocurrency and healthcare before them) off-limits to those without professional degrees and licensure, as well as a budget that includes full-time legal counsel on staff.

In the United States, we still have a window of time to build things that are meaningful and cool.

I don’t know how long it will last. Maybe six months? Maybe one year?

I took two days off to take the pulse of the AI industry, and see what I could learn firsthand. Now it’s time to get back to work.

Last Sunday, I was chatting with other guests at a friend’s rooftop birthday gathering. One woman proudly told us she worked at a local hospital, adding “I am so glad it’s not a Christian hospital!”

I was floored.

People have all kinds of reasons to distrust Christianity. I get that. But a faith-based health organization, Adventist Health of Portland, together with my doctor and the electronic chart system that kept me in direct contact with my care team, was responsible for me being physically well enough to attend the event in the first place.

I was raised as a liberal Episcopalian and wrote a memoir exploring my relationship with Christianity in my early 30s. These days I am never entirely sure what I believe. One thing I can tell you is that part of following Jesus is a core belief that doing the right thing is more important than money.

The website for Adventist Health Portland includes an explicitly faith-based mission: “Living God’s love by inspiring health, wholeness and hope.”

Issues of fairness and equity in the U.S. healthcare are so widespread that it’s hard to know where to even begin to makes things better. I can’t tell you for sure whether faith-based values were part of the decision matrix that made my surgery possible, but it seems entirely consistent.

What I experienced personally can only be described as a “Room at the Inn” situation.

For several months, I had been getting weekly IV iron infusions for anemia. My iron levels were that low. The cause was fibroids with menorrhagia. Over six years, the condition progressed from minor annoyance to debilitating illness.  In the summer of 2022, I was too weak to ride my bike or go out jogging. I slept 12 hours a day. Once, I had been an avid devotee of the Portland music scene. Now having an active social life or dating was out of the question. It was all I could do to clock in to my remote job and cook and clean for myself.

I had already had one surgery (hysteroscopic myomectomy, unsuccessful) the previous fall. The six benign tumors in my uterus came right back, and the bleeding intensified. I was on a wait list through the Oregon Clinic for a second surgery, but had been told the backlog stretched six months or more. Then, right around the Labor Day holiday, things got worse.

I messaged my doctor via the Patient Portal to let her know:

I had my most severe hemorrhage since July 22 last night… There was blood all over the bathroom floor.

Sorry for the graphic detail but I hope you understand I am not exaggerating. I am taking 10 mg of Norethindrone daily. Up until now that has controlled the bleeding successfully.

I got an immediate reply, requesting that I come into her office for blood work. On that same visit, I was given a surgery date — less than two weeks away. My doctor had put in an emergency request for an operating space and Adventist Hospital had answered that call.

I scrambled to get the time off from my work and find a relative to help me through the recovery process. I got a full hysterectomy, but kept my ovaries. Remarkably, the laparoscopic procedure did not require an overnight stay. The hospital staff was caring and attentive, staying late until they were sure I was safe to leave. Within a week, the pain was mostly gone. Within three weeks, only faint scars were visible.

It took over six months for my hormonal balance to even out. But eventually, I got my life back.

The best analogy I can provide is a stained glass window in a cathedral. The procedure itself was a tangible and genuine miracle of modern science. The fact that different medical communications systems were able to connect to bring about a favorable outcome, just as much so.

No doubt you have heard the expression, “Running out of spoons.” It’s for days when your energy and innate natural life force are depleted, when you struggle to get through the motions of daily life. I remember days like that when I was living with chronic anemia and menorrhagia from fibroids. If I could manage to log six hours of work time, order Pho for a late lunch, and get the aforesaid spoon into the dishwasher, it was a good day!

I’d like to add a new term to the lexicon:

“Running out of forks.”

This is for times when you have tasks in front of you that are a bit more challenging. And yeah, there is a Git reference there as well. I would be so honored if anyone ever chose to fork my projects. It hasn’t happened yet. Ever try eating spaghetti with a spoon? Sometimes, you do need prongs.

The particular reason that I’m low on forks is RSI — also known as Repetitive Strain Injury. It’s a common injury for designers, and also afflicts programmers (or virtually anyone else who puts in long hours with a mouse and keyboard) from time to time. Pain is easy to ignore. Until the point when you just can’t ignore it. Imagine the sensation of a white-hot metal wire pressed into the palm of your hand. It only got that bad a few times, but that’s why I learned to eat and do simple tasks with my left hand as well as my right.

My employer in 2022 tried hard to accommodate my injury. In fact, it went away completely for several months! I was not prepared for the extent to which the hand and arm pain came back after surgery for the fibroids (a completely separate issue). My doctor at the time warned me that fully healing from the surgery would require at least a full year. I’ve not seen medical literature to support this, but it makes perfect sense that while my body was recovering from a major surgical procedure, I would be more susceptible to re-injury. Repairing tissue takes nutrients and energy.

I received a small cash settlement for the RSI injury (equivalent to about three months of pay), which helped. The majority of my time this year has gone into planning and research around a social venture to support adults living with chronic and serious illness. The way in which my life changed completely took me by surprise. We are a mostly invisible group, but there are a lot of us. If you live alone, you are even more vulnerable. Particularly in the post-COVID era.

Regardless, I’m not here to complain or feel sorry for myself. I know exactly why the injury got bad. I was working long hours and not taking breaks, trying to reach a milestone on an old Python project. Which I met.  Now it’s up to me to rest until the pain subsides, and keep fine-tuning my ergonomic setup and exercise routine until I am able to put in something close to a normal working day. I’ll be moving back to Oregon in May.  Until then, my expenses are pretty low.

Have the budget to walk into town about every other day and grab myself a slice of pizza, a cup of coffee, or a beer. Not more than that. Or I lose my runway.

I’m grateful to have this time. Just frustrated I don’t have more forks! So many interesting and challenging tasks await. Opportunities to learn… opportunities to build… dream jobs that I am missing out on applying for…

On the plus side, I can still take walks and read books to my heart’s content. I have my energy back, which is a welcome change from life with severe fibroids, two years ago.

Most tasks are fine in moderation. I just have to be extremely aware of when I’m getting to the “red zone” and force myself to stop. I’m learning to be more strategic about where I put my time and resources. I can still do management, customer support, and coaching work until the cows come home! Too bad those first two categories are shedding so many jobs. Keyboard is ok up to a point. Anything that involves a lot of repetitive mouse work (which can include IDE’s) is a a greater concern. If I never open up another Figma file in my life, I’d be ok with that. Design comes naturally and easily to me and I love the interdisciplinary nature of UX work. By contrast, AI very much requires the command line and it’s where the innovation is happening right now in our industry.

I think in the long run, it will work out. I just wish voice recognition software were a teensy bit better. What was missing (last time I checked) was something equivalent to a sudo or vi mode where it was possible to switch from dictation to navigating with the cursor to fix mistakes.  If all else fails, that may be an area where I could make some contributions.

Figured out some hypothetical voice dictation training exercises which would not bore me to tears. The last time I made a chess reference on this blog was almost seven years ago. I’m in a very different place today, but not necessarily a worse one. (Cannot overemphasize the importance of actually getting that busted uterus removed!) Maybe dictating chess moves would be a good place to start for teaching the computer some new conventions for voice recognition.

And other people would benefit too.

So yeah, the same basic principle espoused in that post continues to hold true. I’ll figure things out.

This paper is short and readable, compared to many in the field. It also gave me a better sense of scale in our industry — learned that the cost to train GPT-3 ran to millions of dollars.

The ability to incorporate open source packages into a piece of software allows developers to easily add new functionality. This kind of modular reuse of subcomponents is currently rare in machine learning models. In a future where continuously improved and backward-compatible models are commonplace, it might be possible to greatly improve the modularity of machine learning models. For example, a core “natural language understanding” model could be augmented with an input module that allows it to process a text in a new language, a “retrieval” module that allows it to look up information in Wikipedia, or a “generation” output module that allows it to conditionally generate text. Including a shared library in a software project is made significantly easier by package managers, which allow a piece of software to specify which libraries it relies on. Modularized machine learning models could also benefit from a system for specifying that a model relies on specific subcomponents. If a well-defined semantic versioning system is carefully followed, models could further specify which version of their dependencies they are compatible with.

But mostly it left me wondering, “how and when?”