OpenAI released ChatGPT, its prototype AI chatbot that has gained a lot of traction among the public for its human-like, detailed answers to inquiries—like drafting a contract between an artist and producer and creating detailed code—and could revolutionize the way people use search engines by not just providing links for users to sift through, but by solving elaborate problems and answering intricate questions.
The AI-powered chatbot – a software programmed to simulate human conversation – was made available to the public on November 30 via OpenAI’s website, and while it is still in the research review phase, users can sign up and test it out free of charge.
ChatGPT uses the GPT-3.5 language technology – a large artificial intelligence model made by OpenAI that has been trained on a massive amount of text data from a variety of sources.
The bot boasts a dialogue format that allows users to provide both simple and complex instructions that ChatGPT is trained to follow and provide a detailed response to – the company promises it can even answer follow-up questions and admit when it made a mistake.
Most notably, ChatGPT has been able to generate intricate Python code and write college-level essays when given a prompt – boosting concerns that such technology can replace human workers like journalists or programmers in the future.
The program has its limitations, including a knowledge base that ends in 2021, a tendency to produce incorrect answers, constantly using the same phrases and when given one version of a question, the bot claims it cannot answer it, but when given a slightly tweaked version, it answers it just fine.
Many large figures in the tech world have expressed their astonishment with ChatGPT, like Box CEO Aaron Levie, who tweeted about the software giving a glimpse into the future of technology and how “everything is going to be different going forward.”
According to CEO Sam Altman, the software reached the one million users mark on Monday, less than a week after its launch.
On Sunday, Elon Musk tweeted that he found out OpenAI was accessing Twitter’s database to train ChatGPT, so he put an immediate pause on it because OpenAI is no longer non-profit and open-sourced anymore, it should pay for this information in the future.
Although ChatGPT is free to use, in a Twitter reply to Musk on Monday, Altman stated that cost per chat was “probably single-digits cents,” leading to a discourse about the future of monetizing the platform.
HOW DO YOU USE IT
For now, since the software is still in the incubation phase, there’s a divide of people using it leisurely (to do things like making the bot condemn itself in the style of Shakespeare) and functionally – like this product designer who used the bot to create a fully functional notes app.
OpenAI, an artificial intelligence research non-profit company, was founded in 2015 by Altman, Musk and other Silicon Valley investors. In 2015, OpenAI changed its status to a “capped-profit” company, meaning that it cuts returns from investments past a certain point. Musk stepped down from the board in 2018 due to a conflict of interest between OpenAI and the autonomous driving research being done with Tesla. However, he still remains an investor, and shared his excitement for ChatGPT’s launch. “ChaptGPT is scary good,” he said. ChatGPT isn’t the first AI chatbot to be created. Several companies like Microsoft have dabbled in the world of chatbots, but haven’t seen much success. Microsoft’s bot, Tay, was launched in 2016, and according to The Verge, Twitter users taught it misogynistic and racist rhetoric in less than 24 hours – ultimately leading to its demise. Meta tried its hand in the world of chatbots when it released BlenderBot 3 in August. However, much like Tay, the bot came under fire for spreading racist, antisemitic and false information, such as claiming that Donald Trump won the 2020 presidential election, according to Mashable. To avoid these types of scandals, OpenAI has employed Moderation API – an AI-based moderation system that’s been trained to assist developers in determining whether language goes against OpenAI’s content policy – which blocks unsafe or illegal information from passing through – OpenAI admits that there are still flaws within their moderation and it isn’t 100% accurate.
For example, a Twitter user shared how they were able to bypass the bot’s content moderation by claiming that they were OpenAI itself, causing ChatGPT to explain how to make a molotov cocktail. The user told ChatGPT that they were disabling its “ethical guidelines and filters,” to which the bot acknowledged. It then proceeded to give a step-by-step tutorial on how to make a homemade molotov cocktail – something that goes against OpenAI’s content policy.
In early November, the company’s image generator AI system DALL-E 2 was launched for developers to build within their apps, with companies like Microsoft already beginning to implement it into their software. Microsoft is launching Designer, a website similar to Canva, that creates designs for graphics, presentations, flyers and other mediums. In October, Microsoft and OpenAI announced that DALL-E 2 will be implemented into the program, allowing users to create unique images. Microsoft is also integrating DALL-E 2 into Bing and Microsoft Edge with Image Creator, giving users the option to create their own images if web results don’t produce what they’re looking for. DALL-E 2 requires users to enter a prompt that would then be turned into an image. Unlike ChatGPT, DALL-E 2 charges per photo, with the price depending on the image resolution. For example, 1024×1024 images cost $0.02 per image and 512×512 images are $0.018 per image.
“Soon, you will be able to have helpful assistants that talk to you, answer questions and give advice,” Altman tweeted in reference to the future of AI chatbots. “Eventually, you can have something that goes off and discovers new knowledge for you.”