This Week in Australian Startups - Issue #20, 12th May 2023
Haris Qureshi
GTM Leader | Sharing my thoughts on tech at news.harisq.com | Investing with Cut Through Angels
A leaked internal document from a Google researcher claims that Open Source AI will overtake Google and OpenAI. You can read the?full document here, but some of the key arguments they make are;
The CTO of Hugging Face shared the below graph a few weeks ago as well, demonstrating there will come a natural inflection point where open source models will overtake closed source models.
Whilst this may be true, we know all too well it’s not often the best product that wins. It’s the product that has the best real world value, benefit to users and GTM that wins more often than not.
Ben Thompson?writes;
The biggest pushback I have to the memo, though, is that any determination of a moat is inextricably tied to the business model a particular bit of technology is attached to, which is to say that any sort of blanket statement misses important nuance. Take Google itself: the general way in which its search engine works is understood, but small distinctions in quality still matter; it also matters that Google is the default almost everywhere, that it gets much more user data, has such a huge advertising base, etc. It may be the case that in a few years?we look at ChatGPT in a similar way: yes, its capabilities may be generally matched, but it has small differences that help, it has more user data, a larger and more diverse plugin ecosystem, etc.
There is a bigger push to Open Source happening,?Meta open-sourced a multisensory AI model that combines six types of data. This is significant and a really interesting play by Meta which is contrary to that of the major cloud providers Azure, GCP and AWS.
Majority of?AI products on the market today?are simply interfaces to ChatGPT, and increasingly tech companies are looking at how they can leverage AI.
There’s a real trade-off for businesses to use models like GPT - they don't own the real IP or AI. When they build any products on these out of the box AI solutions they are effectively just building a nice UI on top of GPT. Every time OpenAI updates their model they may need to update their UI, and there’s no guarantee that OpenAI will be updating its model based on any businesses individual requirements.
This starts to pose a real potential risk. Not only is your “AI” product not really AI nor your product, but there’s a real vendor lock in risk and ultimately no differentiation in the market. Take Notion as an example using OpenAI's ChatGPT model - the only differentiation is that you use the tools within Notion and not OpenAI, and the specific context that provides.
As OpenAI continues to invest in ChatGPT, build it’s plug-in ecosystem why would I pay for NotionAI when someone will create a plugin for Notion into ChatGPT making it more useful as I’d likely be able to connect it to many other tools I am using and pay for AI subscription.
For companies really interested in actually building AI and not just reselling it, this is where adoption of open sourced models can start to look more interesting. Whilst they may not be as capable as Bard or ChatGPT from the start, you can have your own teams fine tune and iterate based on your own businesses requirements - and start to build real IP that will build a moat around your business and improve its valuation.
There’s going to value in both approaches as there is with most technology that powers the web today. Many businesses choose to pay for closed source proprietary software whilst others choose to use open source technologies.
领英推荐
Top News
Australia
Around the World
Australian Funding Rounds
International Funding Highlights
If you enjoyed this week's edition I'd love to know, and make sure to subscribe here on LinkedIn or?Substack?to get weekly updates.
Account Executive @ CloudBees | DevOps / DevSecOps / AI Specialist
1 年Great insight Haris