Wow, it’s hard to believe it has already been two years since GrailForge was conceived. It hasn’t been without its challenges but on the whole, things have gone far better than I could have hoped.
The main challenges so far
By far the biggest challenge has been finding reliable financial service providers. While managing expenses was relatively straightforward for a lean startup, securing banking services in Sweden proved arduous. After disappointing experiences with banks living up to all of the worst banking stereotypes, we finally found a reliable business partner in Swedbank.
Accounting proved the bigger challenge. I initially handled it myself so as to better understand what was involved before outsourcing it. Swedish business accounting differed significantly from my previous ventures. I eventually felt I had the essentials down, but not before an unpleasant bait-and-switch experience with a highly unethical accounting software platform.
Though I initially hoped to hire an accountant directly, I ended up settling on an accounting firm. To cut a (very) long story short, I was being outright ripped off and I caught them red-handed. “Scam” is perhaps too strong a word, but chalking it up to mere incompetence does not reflect the ugly reality of the situation.
There are some challenges where you grow stronger for having overcome them and you’re ultimately thankful for the experience. Not so in this case, however I am immensely thankful that we finally have ideal banking and accounting partners. No more headaches - we just pay the bills monthly and everything runs smoothly, freeing us to focus on GrailForge’s bigger ambitions.
In terms of daily operations, the only real hiccup was the result of a Microsoft accounting error. I can’t go into details out of respect for the client’s privacy but the scare came from the possibility of over 2 million Swedish kronor in unanticipated expenses which would have fallen on GrailForge and likely ended us. Only thanks to some intense forensic investigation and having kept meticulous records were we able to prove the fault on Microsoft’s side. It goes to show you’re never too big to be immune to sloppy accounting, and it makes me even more appreciative to have a reliable accountant at last.
Pondering the future
GrailForge was started in the middle of the COVID-19 madness that swept most of the world by storm. Now that the travel restrictions and lockdowns are over, I need to spend some time in Australia to address some personal matters. Once our remaining contractual obligations to our current clients are wrapped up, most of our operations will be paused for a while. How long exactly is uncertain but the GrailForge office lease has been terminated, the office equipment and furnishings are going into storage for now and we won’t be accepting any new customers until there has been some time to work out exactly what direction we want to go in next.
Looking at technology trends over the past 2 years since we started, one thing is overwhelmingly clear. Humanity is standing on the precipice of a huge artificial intelligence revolution. It may not be as heavily pumped as the relatively recent blockchain hype train but the impacts are going to be far more pervasive and life-altering than Dogecoins and NFTs, and it’s going to make much of yesterday’s sci-fi seem decidedly unambitious. The buzz around OpenAI’s ChatGPT is driving mainstream awareness and buzz around AI for tech-savvy and layman alike, but it’s just the tip of the AI iceberg.
A large part of the vision behind GrailForge has always been contributing to a free and open future for all. Many are undoubtedly looking at the tech dominance Amazon has achieved largely off their cloud infrastructure and will be looking for opportunities to achieve the same success in the imminent AI arms race. Up until the waves OpenAI’s latest releases have made in the tech landscape, the future of AI has looked to be likely dominated by the same Big Tech behemoths that dominate other spaces. Amazon, Microsoft, Google/Alphabet, IBM, Facebook. Some lesser but still significant players like Apple, SalesForce, and Tesla. Maybe the Chinese companies behind applications like WeChat and TikTok have the funding and reach to match the pace of Western tech giants’ innovation in those spaces. Why this is so dangerous gets quite philosophical and touches on ethical questions that I don’t think have a place on the company blog. For now, it’s enough to say that I’m reconsidering some of the goals for GrailForge as a result.
The academic and open-source communities around AI are showing a lot of promise. While I maintain little optimism that OpenAI will remain “open” and independent for long in the face of ChatGPT’s recent success, Hugging Face is one of the most exciting tech communities I’ve seen in… well… perhaps ever. The flip side of seeing all this amazing innovation pushing beyond the hypothetical is getting a sense of how redundant many things will soon become. For example, DALL-E and similar lesser-known services like Midjourney already look poised to put everything from stock photography to freelance illustration work into an early grave in much the same way Spotify ended the seasonal compilation CD. Some results may look ‘funny’ now but if you look at the progression of their models from only 12 months ago to now, I would be extremely surprised if we don’t have photorealistic text-to-image generation publicly available within the next 12 months at most, and even text-to-video not far beyond that.
Penguins generated by DALL-E in 2021 (source: OpenAI)
DALL-E in 2022: “A penguin listening to music while programming” (source: Greg Brockman)
It begs the question: is it even worth continuing with some goals or is it better to accept the sunken cost and move on?
Along similar lines, will small-scale audio production even be relevant in 12 months’ time? Is it even worth the expense of maintaining a sound studio when you will be able to get 80-90% of the quality from an AI song generator and an AI mastering plugin? It may sound like a pipe dream now but many of the open models on Hugging Face already show potential, and that’s not considering the amount of proprietary research taking place behind closed doors backed by serious funding.
Even software development isn’t safe. I don’t believe programmers are going to be extinct overnight, but AI will absolutely revolutionise the software development industry within the next 12 months. Probably not to the point that developers disappear entirely, but my prediction is that it will start with the ‘junior developer’ role being heavily devalued to the point that it’s seen as on par with grocery store cashiers. Mid- and senior-level developers will increasingly rely on AI to solve their problems and increasingly become dependent on AI ‘auto-completing’ their code much in the same way that many people let their phone auto-complete entire conversations for them. Eventually, someone will work out how to cross-pollinate the various AI models and feed them back into themselves at which point autonomous AI will eliminate the overwhelming majority of developer roles.
What that means for 2023
Long before Terminator brought the concept of Skynet into the mainstream consciousness, there have been doomsayers predicting the end of civilisation at the hands of our own technological advances. I don’t think the most dystopian scenarios are going to play out, nor do I think artificial general intelligence is an inevitability in our lifetime. I do think it’s undeniable that an unprecedented civilisational revolution is just around the corner though.
For at least the first half of the year, 2023 will be about reading the tea leaves and trying to work out where we fit into that future. Which intended business areas are still worth pursuing and which ones are no longer justifiable. And how GrailForge can best position itself to do our small part in making sure that imminent revolution is a net positive one.
Comments
Comments are temporarily unavailable.