Bigger, better, faster, stronger- where our AI toolbox can take us

Tech has largely promised to empower humanity to make progress in leaps and bounds at the velocity and march fit for a Daft Punk song. As other areas of technology have slowed in innovation momentum, we’ve seen an upward trend of AI tools and their applications to the taxing problems across all industries today. This leads us to explore how the AI toolbox will dominate our future at each layer of the stack, affect what we can do, and how we think.

So what does that look like? Looking at it from a high level lens, we at Tau see possible AI tool box implementations included at different layers…

1.Application layer- machine learning algorithms and computer vision have allowed a world of possibilities including but not limited to-?

• On an individual level: Personal assistant, automation of menial tasks (emails, letters, schoolwork - yikes - ),?frictionless workflow
• On a product level: matching in platforms, robotic automation, health analytics and predictions, financial analysis

2. Service layer- algorithms built cheaper and faster are quickly being commoditized and offered on a service, where we see these tools can be applied to (not exhaustive)-?

• On a product level: algorithm management, feature engineering, new algorithm generation
• On a system level: layer one platforms of managing algorithm workflows and runtime (currently not robust); full stack optimization:?

chip >> data >> algorithm >> server >> cloud optimization

3. Abstract framework layer - just as we built algorithms to learn from large datasets of observations to solve problems, we in turn can take observations of the progress of algorithms as lessons for problem solving -?

• Approaching generating ideas/problems: While DALL-E has not had nearly as much hype around it as ChatGPT, its capabilities is, for the most part, one of delightful innovation in creativity. A computer’s edge over humans has historically been in its speed of computation and ability to ingest and analyze immense amounts of information in short time frames. Generative AI unlocks a new capability- one that touches not on analytical skills but that of creation. That brings us to two points:??

a. Problems are not always solved linearly: sometimes creativeness yields elegant solutions to problems of analytical nature; similarly, sometimes computational bashing can engender products of a creative nature: ex. DALL-E

b. Visual generation is just as powerful as text generation for decision analysis - we often intuitively want search queries ranging from the stock trend of the S&P and the oddly warm January in New York city to return as visuals/graphs instead of blocks of texts; Most often, tools that create data visualizations are more powerful in aiding decision making use case than an 80 page deep dive analysis. Infographic creation is still difficult for Dall-E, but imagine a mixture of Dall-E capabilities with GPT’s text generation… hmmm

• Not everything is a nail: As seen from the recent wave of generative AI fanfare, there’s been an enormous uptick and unabashed enthusiasm for leveraging Chat GPT’s API to legaltech, fintech, adtech, and healthcare, among many, where founders and investors alike are quick to ask how we apply LLMs. While trends begets innovations, not all problems are best solved wielding the ChatGPT tool- other algorithms may provide a more elegant solution to the problem space, and even yet, perhaps AI may not even be much of a value add in some areas, such as consumer tech. We, at Tau, ask founders to choose their tools wisely, and especially if some problems don’t need the AI toolbox at all.?

The future techstack, reimagined:?

Optimistically, a natural progression of what the continuing revolution of AI will look like involves algorithm driven platforms touching all areas of our lives, interactions, and workflows. Here’s a high level fantasy in the healthcare space:?

At all levels of the AI techstack, exciting times lie ahead, whether from the opportunities that come from scrutinizing the application or services layer, or conversely, from an introspection of how we solve problems inspired by how the progression of AI has solved our problems. If you’re building in the space, feel free to reach out to me on LinkedIn or in comments- happy to grab coffee, riff, and indulge in ideas together :)

Primary author of this article is Sharon Huang . Originally published on “Data Driven Investor .” ? These are purposely short articles focused on practical insights (we call it gl;dr — good length; did read). See here for other such articles. If this article had useful insights for you, comment away and/or give a like on the article and on the Tau Ventures’ LinkedIn page , with due thanks for supporting our work. All opinions expressed here are from the author(s).