Is the Generative AI Hype Inflated?
Business viability is key in any ML context
This gloomy economy and intense flu season make finding inspiration difficult. The lights of the Holiday season are still too far away to cheer us up. My trick (or treat?) to kickstart inspiration is to dig into something machine learning related. Someone asked me the other day about my thoughts on Generative AI. I realized I had things to say on the topic, as I have been following it with curiosuity. So why not also share these thoughts with y’all?
If you are not yet familiar with Generative AI, fear not: in basic terms, it involves categories of machine learning (be it unsupervised or semi-supervised) that generate new or complete content from pieces of existing content. It could be from text, image, video, part of images, etc. The outcome is completely original content, automatically generated, that seems human-made.
When hearing these concepts for the first time, normal reactions include: is this for real? Can machines actually replicate human creativity? Will AI replace what we view as uniquely human - an original thought and the ability to imagine?
Take a look for yourself:
Watch this ~10 sec gif recently posted on Twitter that shows some cool examples of what generative AI can do.
Or, sign up for Dall-E 2, and play around with it yourself, to see its capabilities. Dall-E 2 understands the relationship between images and their text-based descriptions - enter text phrases or words, just like a normal search box, and get a set of AI-generated images back.
Most of the “cool” and hype-generating generative AI projects trending online right now are based on diffusion (ref. stable diffusion). Which is an ML algorithm. It starts from a random state and then gradually changes that state towards an image or output, as it recognizes concepts.
Although the visual effects of text-to-image examples are taking the tech (and VC) world by storm (yes, it is very cool!), one should also consider the “less flashy” areas of text-to-text and code-to-code scenarios of generative AI, for example:
Jasper is a blog or written content generator based on generative AI
Metabob is heading in the direction of generating beautiful code from non-optimal code snippets to auto-fix bugs and speed up developer productivity
Melobytes is generating melodies from song lyrics (I also just have to mention this Europen Schlager generator due to my Nordic origin - surprisingly successful in its task)
The most prominent entity in this movement is of course Open AI. If you haven’t heard about it yet, you probably need to read up on it, to stay a-jour. Their research and mission to truly democratize AI by making it available to any application developer out there is amicable. It is also impressive to see what has been built and evolved in such a short time span. To me, it further validates that the world will change with AI at a pace that humanity has not seen before.
At this time, my mind ponders how humanity can better prepare for this rapid shift and where the best opportunities for positive impact may be.
With this new wave of available advanced AI hitting us, we have only seen the beginning of what this spin of ML technology can do. But would I invest in generative AI algorithm-based startups today? You probably would be helped by reading my previous blog post first, to understand my overall stance on ML in general. In short: I view ML as a technology approach and rather a means to an end, over a core business in itself. Generative AI is no different. For me, it will be all about the application…
A simplified depiction of how I view ML vs business viable ML-fueled innovation, with a few examples to help the understanding.
So, what can generative AI be used for, that elevates business workflows and yet make business sense beyond individual use? Personally, I would of course like business tools helping me to generate great visuals for the feelings or points I am trying to convey in a presentation or keynote. Yet, how often would I use that kind of tool? Maybe once a quarter? I would personally love a blog-generating tool where I can feed in some bullet points and generate a blog with my unique voice in the language generated. I would use it monthly. Perhaps the market of influencers and creators is big enough for some companies in this space to thrive. Probably content marketing could be a huge market, but will quickly emerge into a red ocean - there it will become a competition of who serves the workflows the best and all necessary workflow integration with other marketing tools.
But beyond that I am yet to see a tool that helps on a daily basis and for a business critical operation. Which tool will become so sticky it crosses enterprise organizational borders or helps transform how business is done. I can see a lot of productivity gain in code generation. I am yet to see if generative AI can be used for daily tasks, beyond content creation. I also wonder if generative AI could provide help to line out ideas for complex surgeries, for architecture and city planning perhaps, and probably for the art domain of gaming. The application is key, followed by stickiness, use frequency, and market size. The most impactful application of generative AI is probably yet to come.
So will generative AI replace human creativity? In my view, it will as it stands today assist humans, not replace them. At this moment, the human still has to have the original thought - i.e. type the text into the machine to generate the output. Content creators still have to come up with new topics and viewpoints for blogs, even if in the future I could use a tool to help me draft it, and optimize my time for fine tuning. So I think humans will still add to the equation quite a lot and for some time - with the purpose and application.
In conclusion, I must say I feel very inspired in this area. A lot is still yet to be discovered and realized. I am, however, still looking for a business application for enterprise to blow my mind - business and transformative impact wise. On that note: please hit me up if you know an entrepreneur in this space who is thinking as much about the business as the hype factor.