- OpenAI’s ChatGPT blew up last year, and changed how people think about the future of work.
- But while OpenAI became synonymous with AI, its competition hasn’t sat still.
- Privacy issues and data protection will bring many OpenAI customers to its more niche competitors.
The good news for OpenAI is that it’s essentially become synonymous with the current boom in generative AI.
ChatGPT, the company’s overnight success, has emerged as the gold standard in AI chatbots — forcing its larger competitors, namely Google, scrambling to catch up. OpenAI has kept up that momentum by rapidly improving ChatGPT, both by making it smarter and by signing deals with the likes of Microsoft to bring it to more places.
The bad news is that its success may prove to be a double-edged sword. OpenAI’s first-mover advantage is already diminishing quickly as privacy rules, regulations, and good old-fashioned market competition mean that it’s unlikely to dominate the field forever — leaving a giant opening for competitors like Google and international players like Baidu to make their move.
The heart of the matter is that ChatGPT itself is what you might call a general-purpose chatbot platform: It’s designed to be as helpful as possible to a wide audience that spans industries and lifestyles.
Experts and industry insiders say that in the real world, there’s a real benefit to going deep, building AI models that are extremely good at specific purposes, and built to meet the stringent requirements of international privacy and safety rules.
“I don’t think there will be one master AI model that everyone uses, and I don’t want to work in a world where that is the case,” said Naveen Rao, co-founder of AI company MosaicML.
Which is to say that while OpenAI’s status as a pioneer in the space is well-earned, its path to becoming the next great tech giant is less clear.
OpenAI is facing real challenges
Since the release of ChatGPT, OpenAI faced criticism, with schools forbidding its use to prevent cheating and regulators in countries like Italy banning it entirely out of privacy concerns.
Other AI model makers see a growing opportunity in the concerns around OpenAI, and want to pounce on the opportunity.
Meanwhile, leaders including Elon Musk have called on the industry to slow down AI development. The fact that many don’t want to do so for fear of losing market share proves that OpenAI is facing stiff competition that isn’t waiting to see what comes next.
But OpenAI, while on its way to becoming a giant in the field, will still face some tough competition of its own making.
ChatGPT’s ultimate weakness: it’s not niche enough
ChatGPT’s big selling point is the ability to answer virtually any question. It reads a large volume of data — mainly culled from the wider internet — to find the correct pattern to generate an answer to a query.
Not all industries believe their data should be part of the grand, public experiment that is ChatGPT. This belief was bolstered after a bug leaked ChatGPT conversation histories. Companies like Walmart, Amazon, and even partner Microsoft warned employees not to enter sensitive information on ChatGPT. Amazon, not one to miss a promotional activity, then directed its programmers to use its own AI model called CodeWhisperer.
OpenAI’s determination to dominate the field meant it created a more general AI model that others could add on. In that regard, OpenAI licenses ChatGPT to other organizations. This allows startups and other big brands like partner and investor Microsoft to build on top of its blockbuster product.
However, it also leaves an opening in the market for more specific AI models that organizations in sensitive industries will flock to.
Rao said healthcare and financial services companies want to use powerful AI but want to ensure their data remains in their possession and only train AI with the relevant information.
AI isn’t winner-take-all
Competition in the AI space means more choices, which can spur better innovation, the experts said.
“Diversity in generative AI models enables enterprises to build and deploy applications that align with their unique business needs and proprietary data,” said Kari Briski, vice president for AI Software at NVIDIA. “In the case of large language models, even just one-size model doesn’t fit every scenario.”
Briski said being able to customize to fit specific needs even boosts the value of large language models to businesses. And AI, to flourish further, cannot be made such that one size fits all.
OpenAI may have moved its AI to the larger public first, but its very business model to date eventually means other companies will take advantage and serve those it cannot.
Read the full article here
