The Story of ChatGPT and OpenAI: The Evolution of GPT Models
Explore the fascinating story of OpenAI, from its founding to the evolution of GPT models, updated for GPT-4.
You’ve probably been bombarded with all the news and articles about how ChatGPT is taking the world by storm and how to do this and that to imporve your current workflow. But have you ever wondered who created ChatGPT and how we arrived at this point?
The History of OpenAI
Start as Non-profit
In December 2015, OpenAI was co-founded by Elon Musk, Sam Altman (CEO), Greg Brockman (Chairman and President), Ilya Sutskever, John Schulman, and Wojciech Zaremba.
These founders shared the same idea, that AI has enormous potential to bring about positive change, but they were also aware of the significant risks that needed to be considered. Although Musk departed the organization in 2018, the other co-founders remained dedicated to fulfilling OpenAI’s mission.
Their goal was to promote the development of AI in a way that is safe and beneficial for everyone.
Is the goal still true today? Let’s talk about it later.
“AI is far more dangerous than nukes. So why do we have no regulatory oversight?”
Co-founder of OpenAI — Elon Musk (2018)
Transition to for-profit
In 2019, OpenAI (OpenAI LP) announced the creation of a for-profit entity, OpenAI Inc, with the goal of securing more funding to ramp up their AI research and development efforts. To achieve this goal, they received $1 billion investment from Microsoft, who is now the exclusive provider of cloud computing services to OpenAI.
On top of that, OpenAI has also signed a partnership with Microsoft for an exclusive licensing deal for GPT-3, which is the language model behind the ChatGPT.
Fast forward to today, you have probably heard that Microsoft is investing a further $10 billion in OpenAI.
“Today, we are announcing the third phase of our long-term partnership with OpenAI through a multiyear, multibillion dollar investment to accelerate AI breakthroughs to ensure these benefits are broadly shared with the world.”
Microsoft (Feb, 2023)
Do you start to see the trend?
The Evolution of GPT Models
- 2018: OpenAI releases GPT-1, a language model with 117 million parameters trained on a dataset of approximately 8 million web pages.
- 2019: OpenAI releases GPT-2, a more powerful version of GPT-1 that was trained on a much larger dataset of approximately 40 GB of text and has 1.5 billion parameters.
- 2020: OpenAI releases GPT-3, which is significantly larger and more powerful than GPT-2. It was trained on an even larger dataset of approximately 570 GB of text and has 175 billion parameters.
- 2021: OpenAI’s Codex was released, a language model trained on code with 6 billion parameters and 800 million lines of code, powers GitHub Copilot, an AI-powered code completion tool.
- In 2022, OpenAI released ChatGPT, which you are probably using daily right now. ChatGPT is an updated version of GPT-3, referred to as GPT-3.5. OpenAI has not disclosed the amount of data that was used to train ChatGPT.
- In March 2023, OpenAI released the GPT-3.5-Turbo model API to the public, which powers the popular ChatGPT. This model offers capabilities on par with the most powerful GPT-3 models, but at a significantly lower cost — 90% cheaper, to be exact.
- In March 2023, shortly after the release of the 3.5-turbo model, OpenAI announced the release of GPT-4, its most advanced language model yet. GPT-4 boasts improved reasoning capabilities and greater accuracy in solving complex problems. The API access to GPT-4 is being gradually rolled out to developers on a waitlist, with prioritized access available to those who contribute model evaluations to OpenAI Evals.
What will be left for us humans to do? We better get a move on with Neuralink. — Elon Musk’s tweet after the release of GPT-4
I do believe that OpenAI was founded with good intentions and may still prioritize their values, but with the increasing competition in the market and pressure to stay ahead, it’s likely that OpenAI may transform into another profit-driven machine. In fact, it seems that this process has already begun.
You probably see where I’m going with this.
Are We Doomed?
The exponential advancement of AI capabilities from model to model has raised so many ethical questions, here is my take:
- What if we reach a point where AI becomes reliable enough that we no longer feel the need to fact-check?
- Considering the already over-saturated ads within Bing and Google search results, what would happen if the chatbot results were biased to favor the interests of the companies? How might this impact the information we consume, and what can be done to ensure the accuracy and fairness of search results in the future?
- The depth and originality of one’s ideas and creativity are heavily dependent on their knowledge base. However, with the vast amount of information now readily available, it’s possible that human creativity could decline as we become overly dependent?
- What if AI renders many industries such as education, financial services and many others jobs obsolete? How might this impact the workforce and society as a whole?
- While AI companies are experiencing great financial success, it’s important to remember that the datasets used to train their AI models are a collective creation of our intellectual property. In light of this, what measures can be taken to address potential copyright issues and ensure equitable distribution of the wealth generated by AI technologies?
- What if AI is utilized by malicious individuals? Given that scammers are already using AI chatbots to deceive people.
“ChatGPT and other generative AI tools should be regulated.” — Mira Murati (OpenAI CTO), in a recent interview with TIME.
One of the questions asked by the interviewer during the interview was: “Do you think it’s too early for policymakers and regulators to get involved?”
With other major tech companies jumping on the ChatGPT bandwagon and tons of funding flowing into AI, we better hope that regulators will act fast before this all goes sideways.
Actually, on the same day Illumination published this post, Elon Musk openly criticized OpenAI for becoming a for-profit organization.
Conclusion
It’s a bittersweet feeling for me. On one hand, I’m enjoying the benefits of ChatGPT’s assistance. In fact, I built an app that uses OpenAI’s GPT-3 to automatically generate blog post ideas and content, all hands-free. You can read more about it here. But on the other hand, I’m uncertain about where this AI revolution will ultimately lead us.
Thank you for reading! What are your thoughts about this AI revolution? Please share your opinion in the comment section below.
If you enjoy reading tutorials and stories like these and want to support my writing, please consider signing up to become a Medium member.
Update:
I’ve been working with ChatGPT on app development using Flutter for the past couple of months, despite initially having limited knowledge about Flutter and Dart. Stay tuned for a new tutorial on this topic. Additionally, feel free to check out this app, which is now live on the Google Play Store. Here’s the link: Tingles ASMR.
