9621 Agnes Crossing, Lake Suzanneview, New Mexico Island 84604-9295.
ChatGPT is blowing out a candle today, November 30, 2023. Although the generative AI chatbot was only released into the world a year ago, its impact is already
ChatGPT is blowing out a candle today, November 30, 2023. Although the generative AI chatbot was only released into the world a year ago, its impact is already enormous.
A year ago, a relatively unknown company brought an unknown product to market. OpenAI, known to some by name after the earlier launch of the image generator Dall-E, introduced ChatGPT to the general public. ChatGPT was an accessible version of GPT-3, a Generative pre-trained transformer-Fashion model. Until then, such neural networks were known to experts, while the general public heard the thunder in Cologne.
Even before ChatGPT was made available to the general public in December, ChatGPT generated hype. The chatbot seemed to have an answer to almost every question. The fact that these answers weren’t always correct was secondary to the humanity with which the AI presented them. Soon, users were asking more practical questions: “Write an email with this content,” “What do you know about this prospect,” “Can you explain this concept clearly?” or “Would you like to translate this email?” ”
In no time, ChatGPT has become a trusted (but not always reliable) colleague for employees in all sorts of roles around the world. The hype train had begun.
Today we are exactly one year later and the IT world has changed fundamentally. No company in the world dares to make a product announcement without first marinating it in an AI sauce. For good reason, says Manuvir Das, VP Enterprise Compute at Nvidia. “A year ago the world really changed completely. Previously, some companies considered implementing AI technology for specific solutions, but since ChatGPT, there is a universal use case for AI. No matter what industry you are in and what job you do, there is a way AI can help you.”
ChatGPT has shown what is possible and everyone wants to benefit from it. Some challenges remain: AI models must be more accurate and integrate company data as securely as possible. This requires training new models. Numerous generative AI models made headlines last year.
The result is visible. A year ago, companies like AWS, Oracle, HPE, Dell, Intel, Microsoft, Salesforce and Snowflake were telling their own stories. Today they are all writing a chapter in the big AI book. Editor-in-Chief of this book: Jensen Huang from Nvidia.
This is by no means a complete list, and none of these announcements and products would have seen the light of day without the launch of ChatGPT last year.
OpenAI has not been idle in the meantime. The company’s models are still the most advanced at the moment, not least because the company used almost the entire Internet without permission as a data set to train its model. After GPT 3 came GPT 3.5 and 3.5 Turbo. We now have GPT 4 and GPT 4 Turbo, which have more knowledge than GPT 3 last year, are much smarter, hallucinate less and argue more coherently.
On the imaging side, Dall-E 2 was replaced by Dall-E 3 to obtain much more realistic images. This feature has now been introduced in ChatGPT. The tool also has analytics capabilities and can process and summarize documents containing dozens of pages. ChatGPT will even have siblings optimized for a specific context.
OpenAI has now commercialized its offering with ChatGPT Plus and Enterprise. Microsoft, OpenAI’s largest shareholder, has integrated the features of the GPT models into Windows and Bing. And Nvidia had to buy a new safe because OpenAI and Microsoft also rely on Nvidia hardware to train their models.
This training turned out to be an expensive affair. Microsoft has invested billions in OpenAI and built its own (Nvidia-based) supercomputer in the cloud to train the GPT models. It is now clear that not only training but also inference (use) requires a lot of computing power. The ecological impact of the AI boom is therefore not small and will not change immediately.
The investments in Microsoft ultimately earned the company a seat on OpenAI’s board, but the company actually had no say last year. This was felt very clearly when CEO and founder Sam Altman was suddenly fired by the board last month, to the surprise and dismay of the man himself, investors and almost the entire workforce.
Comparisons to Steve Jobs were common: the man who brought AI into the world was pushed into the background by his own company. Just like Jobs, a comeback was inevitable, and with these musical chairs, Microsoft gained a little more control. A year after Altman (IT) became world famous, he became CEO of his company again.
A year after its launch, ChatGPT has developed into a relatively mature product with a commercially interesting offering. More important is the trend that OpenAI has started: new generative AI models are popping up left and right, base models (which are pre-trained and can then be fine-tuned by the user) enable enterprise-level AI, and AI is finding turnkey integration into popular SaaS products .
There is one big winner: Nvidia. All innovations, from OpenAI’s first GPTs to turnkey hardware that allows companies to train their own models today, now run on Nvidia accelerators. The Hopper chips were already popular before ChatGPT, but today they are rolling off the assembly line in larger quantities than ever before, Nvidia is making 10,000 euros in profit per chip and the things are being sold a year before actual production. On ChatGPT’s first anniversary, Jensen Huang will be happy to present the cake.
How does ChatGPT see its own birthday? Who better to request this than ChatGPT itself? The GPT-4 model is “aware” of its birthday and can provide some insights about it. First we asked for an illustration of how ChatGPT would celebrate its birthday:
We then asked ChatGPT to take stock. What does the technology look like in the first year?
As I look back on the first year since my founding, a few key themes stand out:
This year has truly been a journey of growth, collaboration and exploration in the field of AI.
A year has changed a lot. It’s clear that we are somewhere at the top of the Gartner hype curve. It’s hard to predict to what extent the hype will die down next year. The consensus seems to be that AI will certainly continue to grow in importance.
Source: IT Daily
As an experienced journalist and author, Mary has been reporting on the latest news and trends for over 5 years. With a passion for uncovering the stories behind the headlines, Mary has earned a reputation as a trusted voice in the world of journalism. Her writing style is insightful, engaging and thought-provoking, as she takes a deep dive into the most pressing issues of our time.
