Check out the on-demand sessions from the Low-Code/No-Code Summit to learn how to successfully innovate and achieve efficiency by upskilling and scaling citizen developers. Watch now.
In the wake of last week’s release of the DALL-E API, a crowd of startups is sure to follow, racing to build generative AI for the enterprise.
I thought of this as I watched coverage of 50,000 runners converging on New York City for its annual marathon yesterday. It reminded me of OpenAI’s announcement last week about its Converge program, which will provide 10 early-stage startups with $1 million each and early access to its systems.
“I can’t think of a more interesting time to start a startup in recent memory,” said OpenAI Sam Altman in a tweet about the program.
That announcement came just a day before the company released the hotly anticipated DALL-E API in public beta, which means developers can now integrate DALL-E directly into their apps and products — including many that will likely be used for a host of enterprise use cases.
Intelligent Security Summit
Learn the critical role of AI & ML in cybersecurity and industry specific case studies on December 8. Register for your free pass today.
That effort has clearly already begun: While the term “generative AI” for content generation is not new – a quick Google search unearths articles from as far back as 2017 and pre-Transformer models such as GANS (generative adversarial networks) were introduced in 2014 – the term is already taking over my inbox.
There has been a massive uptick in PR pitches related to generative AI in my VentureBeat email over the past 2-3 months. And when I received a missive last week with the subject line “generative AI for plumbers,” I knew the race — and the hype surrounding it — was just beginning.
Expect more startups to get into the generative AI game
Last week, I spoke to Forrester AI/ML analyst Rowan Curran about Forrester’s 2023 Predictions Report. He said he thinks the DALL-E API will open up a variety of new generative AI use cases.
“I’m expecting more startups to get into the game using the DALL-E API, and that in turn will drive both enterprises, especially innovation teams, to use it,” he said. In addition, he expects to see more enterprise-level research and usage in terms of adopting and fine-tuning other large language models for various text and image use cases.
“I think the ability to take the large language models and add this fine-tuning layer on top for some specific industries is where it’s going to start to be very game-changing,” he said. “Construction is a good example – being able to understand, summarize and provide insights from contracts, or on the visual side being able to correctly identify cracks in concrete or stressed iron or something like that, or to generate large sets of synthetic data of damaged concrete.”
Curran added that he thinks “we are just stumbling into the beginnings of this,” and that large language models are actually just one approach.
“There are other types of neural networks and other approaches to developing intelligence, such as using causal reasoning,” he pointed out. He says the promise of tools in areas such as generative AI will likely accelerate “at even more exponential rates than it already is, so it’s tremendously exciting.”
The debate over open source, ethics will continue
The release of the DALL-E API, however, certainly won’t quiet the general debate or criticism around generative AI.
For example, ML community newsletter The Sequence posted an editorial yesterday that highlighted the debate between the API model of generative AI — that is, DALL-E — and the open source model used by Stability AI and its Stable Diffusion text-to-image generator.
“The friction between controlled API versus fully open-source distribution mechanisms will likely be at the center of the generative AI debate for the next few years,” the editorial said. API models enable greater control and simpler mechanisms to enforce fair and ethical behavior of the model, but that also guarantees power remains in the hands of Big Tech. Open-sourcing, on the other hand, “removes the control barriers and enables broader levels of customization and fine-tuning but also opens the door to the unethical use of these models.”
This update included no legal context for the change. It simply stated that it “is possible due to improvements to our safety systems which minimize the ability to generate content that violates our content policy.”
The generative AI race is on — but how it plays out on the way to the finish line remains to be seen.
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.