Rarely has a company had a more spectacular breakthrough in the past year than OpenAI. Now comes the most difficult phase: confirmation. What will 2024 bring for OpenAI?

2023 can without a doubt be called the year of OpenAI. In just one year, the company grew from a humble startup to one of the most talked-about organizations today. OpenAI is almost single-handedly responsible for the current hype around generative AI.

The success puts OpenAI in a completely new situation for the company. Breakthrough is one thing, but confirmation must follow in 2024 and many promising startups have stumbled upon it in the past. Can OpenAI rise to the top of the technology world or will it just as quickly fade into obscurity? The company’s fate largely depends on how the hype around generative AI develops.

Unexpected success

The successful year 2023 actually began for OpenAI at the end of 2022 with the launch of an application that no longer needs introduction: ChatGPT. The chatbot achieved success that OpenAI might never have dreamed of. After just five days, one million users had signed up and by January ChatGPT had 100 million users. ChatGPT broke all launch records: it took Facebook more than four years to attract the same number of users. To mark ChatGPT’s first anniversary, we took a closer look at the unprecedented hype.

The launch of ChatGPT completely changed the status of OpenAI. The company had existed since 2015, but in the early years it was more of a hobby project for Elon Musk. Three years after it was founded, the foundation for today’s success was laid with initial research work on a generative pre-trained transformer, GPT for short. This was followed by GPT-2, GPT-3 and finally GPT-3.5, on which ChatGPT is built. We are already at GPT-4 turbo The fifth generation could follow next year. With the Dall-E image generator, OpenAI has another popular generative AI tool.

The rise of ChatGPT also made CEO Sam Altman a “tech celebrity.” While a year ago few had heard of him, now everyone knows who Altman is. He traveled around the world last year and was even invited to the White House for coffee. Altman is sometimes referred to by the media as “the new Zuckerberg.”

Takeover failed

It didn’t matter that ChatGPT’s birthday was a small one. On November 17, Altman announced the surprising news that he would be leaving OpenAI immediately. Altman and the board no longer saw eye to eye. A consequence of OpenAI’s special organizational structure.

OpenAI is a commercial company run by a non-profit organization. In other words: OpenAI needs to make a profit, but it actually doesn’t. Altman found himself in the profits camp, looking for ways the company could capitalize on the generative AI jackpot. OpenAI has, among other things, introduced paid ChatGPT subscriptions and APIs for developers to offset the high cost of upgrading GPT models. This led to increasing resentment between the CEO and the board, who felt that Altman was being too reckless and not sufficiently transparent when introducing new features.

After a power play that lasted just a few days and even saw an employee revolt brewing, Altman prevailed. In a Steve Jobs-like comeback, he returned as CEO to draw comparisons with another tech industry giant. The rebellious board was allowed to resign and was replaced by new board members who were more sympathetic to Altman. OpenAI showed initial success, but Altman cleverly used the matter to his advantage.

The rest cannot follow

The competition will undoubtedly have hoped for a different outcome. After ChatGPT, every company suddenly had to do something with generative AI. Technology conferences were almost exclusively about generative AI, and there didn’t seem to be a word that couldn’t be followed by the suffix “GPT.” Sometimes they are effectively innovative applications, but just as often AI-related announcements felt like shameless jumping on the bandwagon.

Google par excellence has made desperate but unsuccessful attempts to surpass ChatGPT. With DeepMind, the technology giant had been at the forefront of AI research for years, but was now suddenly at the cutting edge. Bard was born in February, but the launch went completely wrong. Google has made mistakes that we are not used to, a sign that the company no longer considers itself untouchable.

With Gemini, Google hopes to have found the right answer to be competitive again in 2024. The announcement panned benchmarks where Gemini outperforms GPT-4 in all areas. These only apply to the Pro version for developers: the standard version of Gemini doesn’t seem to be any better than GPT-3.5. When GPT-5 sees the light of day in 2024, all of these benchmarks could be wiped off the table again. So Google doesn’t have more than one address left for the time being.

AWS watched in the background for a long time, but during its re:Invent conference at the end of the year it finally gave Amazon Q a try. Despite all the superlatives that AWS attributes to its chatbot, it is nothing more than an attempt to keep up.

The only potential competitor that will really scare OpenAI at this point is Anthropic. This start-up is barely two years old, but is already making a splash with its ChatGPT counterpart Claude. Anthropic hasn’t achieved major commercial success yet, but that could come soon as AWS and Google outsmart the company. Fun fact: OpenAI’s board tried to poach Anthropic CEO David Amodei as Altman’s replacement.

The influence of Microsoft

We deliberately did not mention one party in the overview above: Microsoft. Redmond takes a different approach and pays OpenAI with money. An initial investment of $1 billion was expanded by another $10 billion this year. Microsoft now owns 49 percent of the shares.

Thanks to these investments, Microsoft receives exclusive priority over OpenAI technology and makes good use of it. With Bing Chat and later Copilot, Microsoft is becoming a serious challenger in the AI ​​industry. This sometimes creates tensions with OpenAI, which is both a partner and a competitor.

The unrest at OpenAI wasn’t bad for Microsoft in this regard. Microsoft also immediately sided with Sam Altman and initially offered him a job, but then did not put any obstacles in the way of his return. We’ll never officially know what role Microsoft played in this, but the fact that it has now been given a seat on the board suggests that Microsoft is more than a generous sponsor. A development that regulators will view with caution.

Killjoy Europe

It’s the downside of OpenAI’s success: public attention also means that the company is more often in the crosshairs of politicians and regulators. OpenAI has already had to discover this several times in 2023. There was a dispute with European countries over the GDPR law. ChatGPT can also answer personal questions about you, but this information is not always accurate. Data protection authorities don’t like the fact that you as a user have little control over this information.

Altman has also openly spoken out against the upcoming AI Act. It could classify ChatGPT as a “high risk” system and subject it to stricter regulation. Among other things, OpenAI needs to be more transparent about how it collects data and uses it to train models. Altman considers these conditions technically impossible and has even threatened to leave Europe. It reminds us of Zuckerberg’s numerous threats to shut down Facebook and Instagram in Europe because he doesn’t like their data protection laws. Ultimately, OpenAI, like Meta, will have to conclude that it is not above the rules.

There are also opponents in our own country. As a belated Christmas present, OpenAI received a copyright infringement lawsuit from the New York Times newspaper. This isn’t the first AI tool to get into trouble because of this. Such accusations stem from the fact that companies like OpenAI have scoured the internet to get enough data to train models. The fact that copyrighted content was used is an unpleasant thought that we are now confronted with.

Ultimately, OpenAI, like Meta, will have to conclude that it is not above the rules.

Is the world ready for generative AI?

The question arises: Is the world ready for generative AI? There is a lot of talk about how generative AI will improve our work and personal lives. Generative AI can relieve us of many routine tasks, leaving us more time for the creative work that sets us apart as humans, it seems. At the same time, generative AI continues to raise many questions.

Data protection comes first. Companies are unwilling to share their trade secrets with ChatGPT, which sometimes results in the use of the chatbot being restricted or banned altogether. In addition, not everyone is ready to welcome ChatGPT as a colleague for fear that the chatbot will one day be able to completely take over the tasks. A recent survey by Deloitte shows that AI is far from being integrated into the workplace: almost seven percent of the Belgian population use AI tools in the workplace.

How the generative AI hype will develop further promises to be the technology question of 2024. We’re now in the hype phase of the Gartner adoption cycle: there’s a lot of experimentation going on, but there’s no deeper integration yet. This should now follow in 2024. In this phase, generative AI must deliver on its big promises, otherwise interest will quickly wane. Which way does generative AI go? The fate of OpenAI hangs in the balance.