introduction to this blog

The name is Austin Harshberger and this is my personal blog site. It contains arbitrary thoughts on various topics and information about things I build. While I don’t profess to be an oracle, I do hope there’s some helpful insight in my sharing of what I’ve learned that would be useful for others. If it is indeed useful or not is up to you as The Reader.

about me

I’m a techie located in San Francisco, currently in my early 30s. I accidentally became a self-taught engineer after dropping out of a mostly failed attempt at preparation for law school because I was lucky enough to get noticed at an Apple store I was working at, after creating an automation script, that streamlined store operations using Bash. Apple liked it enough to hire me at 19 with no degree or previous experience as an engineer and moved me to Cupertino with a sign-on bonus. I was there for a few months before deciding it wasn’t for me. Thankfully they didn’t ask for the bonus back. After that, I built some Android apps, ironic purposefully, and took an immersive course in UX design before getting a job doing customer service, with some moonlighting as a developer, at a place called IDEO. I was there for a bit before getting a job at GitHub, where I was long enough to see the Microsoft acquisition. I left GitHub to join the Web3 space, first at Coinbase, then NEAR, Parity Technologies, and finally Aleo where my focus was on zero-knowledge cryptography.

Probably thanks to my Mom, people generally tend to like me so I progressed through my career in a vertical-and-up fashion, from engineering to recruiting, to program management, then product management. I ended my career working for other people as an engineering manager, which I thought I’d enjoy more than I actually did because it was mostly politics and I’m not the best student of machiavellianism, in fact, quite the opposite, unfortunately. Now, thanks to a confluence of events, a good idea, and too many STIs, I am the founder of a startup called status.health focused on improving sexual wellness and reducing STI transmission rates.

ai tools and me

I am an AI tools maxi and believe, strongly, that they help more than they hurt. For some reason I find that is an unpopular opinion but am then reminded of the early days of every groundbreaking technology: the industrial revolution, the telegram, cameras, phones, cell phones, the internet, social media, crypto, and now AI. EVERY SINGLE TIME it scares people. Luckily, I don’t scare easily and am often eager to learn something new.

AI tools are nowhere near as good as humans at any particular task–they are “generally” intelligent but not expert at anything regardless of how good you are at prompting them. They are much cheaper and easier to delegate to than an employee in the context of a startup or as a business owner. For consumers that fact check responses they can be more effective than search engines at following train of thought thereby enabling a “stream of consciousness” Siri and her ilk only wish they could deliver. As such, AI tools have their place in growing a startup, helping with learning, and other things that require specialized knowledge or a huge learning curve: programming, complex math, a new language–you name it. AI is terrible at being creative. Probably not news to most of you reading, but worth reiterating as I continue to find it annoying since I am not the best “creative” either. When making assets for products I build, it’s often the hardest task there is; I guess job security for the designers, artists, and writers out there. Congratulations.

I feel strongly that everyone should and probably will eventually use AI, but I also believe in being clear about when it was used. Why not just say:

❔”I created X with an initial prompt and multiple iterations based on research. After some back-and-forth, Claude and I got X to where it is now and I am proud of it. I probably don’t know everything about it but I know most of it and can answer any questions you have.”

Not attesting to the way X deliverable was created only impacts your reputation negatively when you inevitably get stumped answering a question about something outside your depth. It’s like a teacher calling on you in English class after you chose to spend the night playing World of Warcraft instead of reading “The Scarlet Letter.” You should have just read the thing, but short of that, letting your teacher know you played WoW instead and plan to listen to the other students’ answers to learn that way, is the next best thing.

In short: AI-assisted anything should be attested to (clearly marked), and so should human-generated content because it addresses two key concerns most people have: 1. a person misrepresenting work and not actually knowing what a deliverable is or does, and 2. AI’s reputation for being a human-killer vs. a human-enabler, except, of course, for the creatives. Assigning credit where credit is due is not new. It’s much like when that same English teacher insisted on annoyingly formatted in-line citations or Chicago style for book reports.

To make attesting to AI-assisted work vs. that of human work easier for myself and my businesses, Claude and I created a site called attest.ink which allows automated badge generation. I use attest.ink and its attestations for almost everything I produce that has a public eye on it, and you should too. It’s free! Check it out and let me know if you have any questions or suggested improvements: info@attest.ink.

what badge attestations look like

Human Generated AI Assisted AI Assisted AI Assisted AI Assisted AI Assisted

ai and this blog site

Blog posts on this site are not AI-assisted other than for grammar and spelling checks because I am terrible at both. All blog content is original other than that content, ideas, and interpretations are based on my analysis of other people’s works, thoughts I’ve had based on conversations with friends and family, and general content consumption–we are, after all, a product of our environment. Not dissimilar from how AI is a product of the data it has access to. While I could easily use Claude to write blog posts though ChatGPT is better, I think it’s more valuable to write them myself. Doing so serves as a good reminder of the time it takes to make something without AI-assistance, and it’s fun despite how grueling I find long-form writing. I will admit that I am also creating this blog and all its posts so I can train the various AIs on my writing style, voice, and humor, since they are currently woefully bad at both due to the lack of data out there representing me. The non-blog content on this site such as the site itself, was created with AI assistance because Claude is a faster engineer than I am. That said, you can take solace in the fact that you are reading my thoughts, pure, cluttered, and sporadic as they may be. Look out for the “Human Generated” badge on each post.

mostly open source

For the most part everything I create is under the typical open source MIT License. You should feel free to use my words, my website content, and thoughts, I just ask for proper citation, e.g., Harshberger, A, $DATE, $BLOG-POST-NAME or attest.ink, A. H. $DATE. The exception to this rule is anything over at status.health or where I explicitly mark something under copyright with ©. In those instances, just shoot me an email for permission or with any questions: x@97115104.com.

for those that like to internet stalk

You can find more info with a DuckDuckGo query, on LinkedIn, or the bio on my startup’s website: status.health/team.

contact

Feel free to send through a message via email or LinkedIn or GitHub or Telegram.

‘Twas nice to e-meet you, dear reader!