Point of view

Point of view - GenAI, Startups & VC Investment - Will GenAI Kill Innovation?

Alex Eleftheriadis /
#point-of-view
Feature image

I was recently invited on a panel on “GenAI, Startups & VC Investment” at the wildly successful GenAI Summit. This event is the brainchild of Giorgos Nikoletakis, founder of 100mentors, and has evolved into the premiere event in Southeastern Europe on how companies, universities, and schools integrate new GenAI capabilities into their everyday workflow and R&D operations.

A few days before the event, on February 23, 2024, the Guardian reported that well-known TV and film producer Tyler Perry had paused an $800M expansion of his Atlanta studio complex after the release of OpenAI’s video generator Sora, and warned that a lot of jobs in the film industry will be lost to artificial intelligence. The day before, Nvidia’s founder and CEO Jensen Huang was quoted saying that we should no longer strive to teach our kids to be programmers, since – with AI – everyone can now be a programmer. Their focus should instead be on acquiring domain expertise and upskilling.

There is so much change driven by AI developments that it is very hard to figure out what this all means and where things are headed. A key concept that permeates startups and VCs is “innovation.” Understanding what innovation means in the context of GenAI is key to help us untangle the mess.

Large Language Models (LLMs) such as ChatGPT are trained on massive amounts of data representing what humanity has produced in the past. In that sense, they are trained to produce outputs that we are statistically likely to produce. There is a great quote attributed to André Gide, a French author and 1947 Nobel laureate in Literature: “Everything has been said before. But since nobody listens we have to keep going back and begin all over again”. In some sense, GenAI takes all that has been said before and uses our prompts to “say it all over again,” just to us.

We are all trained extensively at school to write summaries and essays. Creating such content is what we call intellectual labor. A big part of our education is spent on learning how to do this in an authoritative way, using research and critical analysis. We were all amazed to see LLM tools produce such work at the click of a button. We became outright astonished when we saw that AI tools can cross modalities: they can generate images from text descriptions and, recently, video as well. It takes years or training, practice, and lots of talent, to get to do this as a human.

In the past we used technology to engineer ourselves out of mundane manual labor. It seems that we are now engineering ourselves out of mundane intellectual labor. And mundane intellectual labor is abundant in today’s workplace.

Innovation, however, goes beyond the mere ability to create and its output exceeds what can be obtained by mundane intellectual labor. Innovations are unpredictable, often deviate from the norm, and are filtered by quantitative and/or qualitative criteria, scientific or social, before eventually becoming part of normality. You could generate such unpredictable “innovations” using a random process, but it is the “intent” of the innovator that makes the difference in creating one that is likely to be useful or well received. This applies not only to engineering and science, but to art and any other type of intellectual activity.

When it comes to startups and software, the implications of GenAI are already being felt across the board. Simply automating mundane intellectual tasks is no longer a unique selling proposition in the sense that one could readily do that using LLMs and appropriate tool integrations. One of the first questions a VC should ask today of a founder is why what she is building cannot be done using ChatGPT.

Another compelling aspect of GenAI is its use as a co-developer. Software developers in the past used building blocks composed of very large libraries. Learning the architecture of these libraries and gaining experience in their use was a career asset. It was a marketable skill, something that you would list in your resume. Today you can instruct tools like ChatGPT to create a library to your specification, in the language of your choice. You are more of a software architect. In fact, as Nvidia’s Jensen Huang pointed out, you can even go all the way and prompt your way to generate a whole application.

VCs and startups were always about exceptions – outliers. Especially in deep tech, domain knowledge has always been critical. In that sense, the future of VCs and startups is brighter than ever because a lot of the work that a company needs to do, and pay for, includes tons of mundane intellectual tasks. Delegating such tasks to GenAI can leave more resources for what truly differentiates a product.

Companies should therefore make sure that they either use GenAI to accelerate building their product, or provide GenAI features directly to their customers, or both. If they don’t do at least one of the two I am pretty sure they will be left behind.

I was a Ph.D. student when the first web browser was released in 1994. I ended up spending the bulk of my engineering career on figuring out how to best transport digital video across the Internet for real-time communication. Today it is a multibillion-dollar industry that simply did not exist when I was a student. I am sure that similar innovations, and industries, will evolve in this new era and, probably, at an accelerated pace. There are a million questions to answer related to ethical use, lifestyle changes, screen time and education for our children, energy use and the impact to the environment, etc., but societies always had to struggle with such questions with any new tool. We are very fortunate to live in an era where the tools at our disposal got to be so intelligent. We just have to make sure to be intelligent about their use.

PS: In addition to deploying ChatGPT internally at Big Pi for some clever database processing, we used the system to provide us with a quote for a “canned” rejection email. We reviewed several from well-known historical figures but could not converge on one. Here is what ChatGPT responded to the prompt “quote about not giving up after being rejected for funding”:

Remember, rejection is not a reflection of your worth or potential. It’s simply a redirection towards a better opportunity. Keep pushing forward and stay committed to your vision.

  • ChatGPT, 2024