AIFA comments in Financial Times’ Sifted, EU investigation into OpenAI

The Financial Times' Sifted recently contacted the Artificial Intelligence Founders Association (AIFA) for its views on the recent European Commission's investigation into Microsoft/OpenAI and startups' early access partnerships with OpenAI.

Read our president's views below, and click the link below to read the full article @ Sifted.

As founder of Decorte Future Industries & president of the Artificial Intelligence Founders Association (AIFA), I was asked by the Financial Times' Sifted for my thoughts on the European Commission investigating whether some AI startups' early access partnerships with OpenAI are in breach of competition rules (part of its wider investigation into OpenAI and Microsoft).

Here's what I had to say:

👉 It’s normal that an economy, including one of partnerships, starts forming around models such as OpenAI's ChatGPT. Like with any partnership, alongside advantages, such as early access, can also come disadvantages, such as a dependency on reliable access to third-party GPAIs (often based in the US).

👉 While companies like Sana from Sweden may have preferential deals with OpenAI to get early access to GPT4, in order to remain versatile they cannot rely just on that partnership, but also use or have used Anthropic’s and Google’s offerings.

👉 At present, OpenAI’s experimental structure, which no doubt attracts the regulators’ attention, is the only unusual aspect of otherwise normal market behaviour. Even in the completely fictional case where OpenAI tomorrow selects just a single European company as its exclusive partner, this would primarily harm OpenAI’s stated mission, and likely ultimately benefit its competitors.

So, roughly summarised (with thanks to Daniel Steele):

✔ OpenAI do not have a monopoly, the process of creating a GPT is publicly known and there are multiple models to choose from, including currently available OpenAI models.

✔ Given there's no monopoly over the technology, the thing that makes OpenAI's implementation different/sought-after is just the product itself. It's a high-quality and well-implemented product.

✔ Giving earlier access to newer tooling for top-tier customers is a normal thing for businesses to do, especially since it's just early access and the models have historically become publicly available anyway.

⚠ However, OpenAI did start as "Open" AI - the fact that they firmly switched to a for-profit ethos but still have a hybrid for-profit/non-profit structure naturally raises eyebrows, especially considering the power of the product they wield.

Read the full article here