ChatGPT may have never achieved widespread user adoption: "We were very wrong"

ChatGPT is a feature tool in the artificial intelligence space. Considering its immense popularity, it's hard to believe that it's just been a year since its debut. On this occasion, Brad Lightcap, OpenAI's COO, divulged behind-the-scenes details of the launch of this popular tool that internet users are discussing actively today.

The decision to launch the service was not straightforward. The team behind it had lengthy discussions, primarily focusing on one concern: Should ChatGPT be released at all?

OpenAI had a limit to its resources. Its most valuable assets were specialized graphics processors, the core technology that allows virtual entities to converse like another person. The advanced processors' cost reflected their technological sophistication. The company intended to operate mainly in the B2B market — providing services for corporations and developers — without any plans for consumer popularity.

Sam Altman, CEO of OpenAI, who temporarily left the company under volatile circumstances a few weeks ago, stood firm. He believed letting users interact with the tool was the best way to test its effectiveness. This intuition proved accurate.

ChatGPT set all-new user acquisition records. Today, it is actively used by 100 million people and also by 92% of companies listed on the Fortune 500 list, as CNBC points out.

In an interview with CNBC, Brad Lightcap confessed that rolling out ChatGPT to ordinary users initially seemed daunting.

We had to change our perspective and accept that ChatGPT could be easily used by the average person.

Lightcap confessed that the success of another OpenAI tool, DALL-E — which generates images — fueled hope for a positive outcome. This tool enabled image generation based on text descriptions. However, it remained unclear if ChatGPT would evolve into a commonly used tool or just a novel entity that internet users found intriguing.

I recall we wagered on how much ChatGPT would grow. I bidded "at most" —suggesting a million simultaneous users. We were greatly mistaken—he stated. He also highlighted the potential expenses of scaling resources since he oversees finances at OpenAI.

Lightcap further stressed that ChatGPT's success originates from its versatility. Regardless of whether someone works in engineering, chemistry, or biology, there is a vast array of knowledge that exceeds anyone's grasp. Nobody is all-knowing, even if they are daily practitioners in a particular field.

It would seem that Chat GPT's triumph would only bring more achievements. The still-official startup — according to media estimates — could potentially reach capitalization levels of 86 billion dollars.

However, all was not well in paradise. In October, the OpenAI board dismissed the company's founder, Sam Altman, prompting a flurry of resignations and rumors of more to follow. After a few tumultuous days, Altman returned, and Microsoft — an investor of 10 billion dollars in the new tool — secured a seat on the board as an observer (without voting rights).

OpenAI also faced another problem of unhappy users. For a long period, the company offered two service models — a free one, which allowed access to a previous generation chat, and a paid one, offering access to the enhanced model of ChatGPT 4.0 or Turbo model. This newer model is currently only available to those who activated the paid service prior.

There are thousands of people on the waiting list. We hope to accommodate all of them by 2024 - said Lightcap. Yet, the internet is teeming with desperate posts from programmers, journalists, and students pleading for immediate activation of the paid ChatGPT.

OpenAI is dealing with additional challenges. The future development direction of its offered services is still unclear. Altman was criticized by the board for attempting to commercialize the developed solutions prematurely. Accusations also emerged that his communications with the board were insincere.

Research into AGI (Artificial General Intelligence) is also contentious. This more advanced tool provokes various ethical dilemmas. Some scientists and experts take it so seriously that they advocate for rigorous control over AGI research. In May of this year, researchers from the Center for AI Safety appealed to governments, warning that artificial intelligence is a "threat of extinction" comparable to an atomic bomb.

This cautionary sentiment isn't new. In 2016, the late esteemed physicist Stephen Hawking issued a similar warning. He argued that artificial intelligence's aims must align with humanity's. He didn't reference an atomic bomb, but used an anthill analogy, stating that ants wouldn't deter humans from building a dam. Hawking cautioned that humans could become as insignificant as ants to AI.

If we finally manage to create real AI, it will either be the best or the worst thing in human history - Hawking emphasized.

Several scientists and businessmen, including Elon Musk, shared this AI development skepticism.

ChatGPT is an advanced artificial intelligence system designed for communication and interaction with users in a comprehensible and natural manner. Its operation is based on natural language processing algorithms, which enable it to understand and generate text.

ChatGPT, or Generative Pre-trained Transformer, is an AI model that has been trained on vast datasets to generate high-quality text. It incorporates advanced natural language processing and machine learning techniques to comprehend and create texts based on given context.

Operating on a generative model means that AI can produce new texts and images based on provided input. It can answer questions, offer advice, generate descriptions, and even write articles. OpenAI is not the only company with its own language model. However, the first experimental chatbot, created in 1966, could only generate simple sentences made up of pre-set phrases — it could not modify or create anything new.

How intelligent is ChatGPT? For decades, the Turing test — developed 73 years ago — has been the hurdle for any computer system. This test measures a machine's ability to converse using natural language and indirectly establish its capacity for human-like thought. If the system's responses are indistinguishable from a human's, it passes the test.

The question of whether ChatGPT finally passed the famous test is not straightforward. According to scientists, it rather "broke" the test. While it can maintain a seamless conversation, it is still incapable of independent thought.