The AI Buzzword

Is this bad for startups?

Welcome to the 591 new people who have joined us since last Monday!

Join the community of 13,153 individuals who are cutting through the noise by subscribing here today:

Hey friends 👋 ,

Welcome to Through the Noise!

AI has become a mega-buzzword. Historically this is a bad sign for new startup creation.

So how do you build an enduring, differentiated business with AI?

It’s time to strap in and enjoy.

Read time: 3 minutes

Escaping the Buzz

Artificial Intelligence (AI) has become one of the most talked-about topics this year. Many people are quickly jumping on the bandwagon and rebranding their businesses as “AI companies”, regardless of whether or not they are actually using AI in a meaningful way. The mere use of the word AI is now seen as a necessity for highlighting business presence and securing venture capital.

So how do you create real value? By solving a real problem.

I recently watched Greylock General Partner Reid Hoffman's interview with OpenAI CEO Sam Altman. Needless to say, the threadboi inside of me turned my learnings... into a thread:

One part I didn't cover when writing was the startup angle. We'll explore this today.

What's the big picture?

Looking at the large language model (LLM) landscape right now, there are a few big players. OpenAI released GPT-3 to the masses back in 2020; AI21 Labs released Jurassic-1 in August 2021; Cohere released its range of models (small, medium and large) in November 2021. Trying to train your own model from scratch is a tall order for any new entrant. To say the least– the barriers to entry are ridiculously high.

We really shouldn't have a world where every single company is training their own GPT-3, it would be massively environmentally costly, compute costly, and we should be trying to share resources as much as possible.

Aidan Gomez, Co-Founder and CEO of Cohere

What we'll see going forward is a new set of startups that will take a large language model of the future and tune it to create 'the' model for medicine, computing, copywriting etc. These are the companies that will have enduring value because they have a special tuned version of the model.

Altman believes the manipulation and fine tuning of artificial intelligence by human intelligence will generate the most value. These companies won't create the base model. Instead, they develop a unique data flywheel that improves over time, separating themselves independently from the stack. These startups can 'shop around' when a better model appears leading to a more efficient marketplace.

A Little Something Extra

  • 🗞 Department of Product Briefing: I recently stumbled across a product newsletter that doesn't suck. It covers the latest product launches, news and analysis to help you build winning products, read by 16,000+ people from Spotify, Amazon & Apple.

  • 🗞 Psychology of Marketing: My friend Abhi writes a marketing newsletter that's based on science to 12,000+ marketers.

  • 🤝 Sponsorship: If your company would like to partner with me via sponsorship with the newsletter, please fill out this partnership form.

That’s all for today friends!

As always feel free to reply to this email or reach out @thealexbanks as I’d love to hear your feedback.

Thanks for reading and I’ll catch you next Monday.


If you liked this piece, subscribe below: