What do the latest rapid developments in AI mean for start-ups?

Diarra Smith
Ada Ventures
Published in
3 min readFeb 28, 2023

--

AI investor John Spindler on new risks and opportunities

I was delighted to have a chat with John Spindler, CEO of Capital Enterprise and former Co-founder of AI Seed, last week. John has written cheques in over 200 predominantly deep tech/AI startups since 2014. When it comes to AI, he knows what he’s talking about. He is the perfect guest to discuss the biggest story of the year in tech so far: the rapid rise of generative AI and what new tech like ChatGPT means for start-ups.

Recent advances in generative AI mean machines are now capable of creating novel content rather than simply analysing or processing existing data. That could have a huge impact on the opportunities and risks leaders of venture-backed businesses face.

John remains firmly optimistic. This is an abridged version of our fascinating conversation:

Which start-ups will be impacted?

The impact and usefulness of AI software has overall been slightly disappointing … But the 95% of software companies who have been able to ignore it till now will be most impacted by the latest advances in Generative AI particularly the Pre-Trained Large Language Models and their cousins Vision Language Models.

And what about the opportunities?

Every C-suite, B2B operator, director of a large organisation is talking to their CTOs and asking: ‘What are we doing with AI, especially generative AI?’ People will commission much more AI modelling across the board, not just in data generation but classification of data and prediction.

Software innovations will make AI simpler to use. Pre-Trained Large Language Models are generative because we can sample from them, in effect ask them questions using natural language. Because they are simple to use, more people will use AI than ever before. Those who are more aware of it will use it first.

One risk is that technical debt will be greater — software code that has taken months and months to write, deploy and optimise can become superseded and surpassed by code that Generative AI can write in seconds. To keep ahead, relevant startup software engineers will need to work harder and smarter.

Who is going to win?

Large Language Models cost millions of dollars to train. They are expensive to build and expensive to build upon and use. They are not for everyone but Aaron Levie, the CEO of Box remarked that “The latest AI wave is the first time we’ve seen a real “innovator’s dilemma” moment in 15 years. Incumbents get to grapple with technology that may not be ready for all uses, while startups can begin where the tech is “good enough” and iterate over time. The race is on.” John agrees.

John also believes that the best opportunity for start-ups won’t be generating large models but will be working on the application layer and the tooling layer; John agrees with a brilliant take on a16z.com. Applied AI solutions to standing problems is where the excitement will be.

Who will be the losers?

Anyone who tries to ignore it, who won’t spend time on it. John recommends that engineers take a day out and look at how generative AI will impact their ability to change and disrupt their business.

Last piece of advice?

Play with it! Play with ChatGPT, play with Hugging Face, play with Stability AI. Imagination in humans trumps anything machines can do, so immerse yourself in it and play with it. And don’t get intimidated. All change is better for start-ups than for incumbents.

Any good thought leaders on this?

Follow Ethan Mollick (@emollick) to learn more

Thanks to John for his insight and we’ll be continuing to follow his thoughts on a rapidly changing AI landscape.

--

--

Diarra Smith
Ada Ventures

Diarra is an experienced operator having spent nearly 10 years executing best in class support services for founders and directors from UK growth companies.