Many movies have been made about the rise of robots in order to show people what the world might look like in the future. However, with the rise of artificial intelligence, specifically generative AI, we are much closer to knowing. So, what exactly is generative AI? Simply put, generative AI is artificial intelligence that can create new content rather than analyze or act on existing data. Perhaps you've heard more and more about OpenAI and their new product ChatGPT, which has been trending on Twitter and has gained over a million users in only 5 days since its launch. But what exactly is ChatGPT? What exactly is OpenAI? And how does this game-changing technology actually work?
Generative AI has been described as the most talked about technology in Silicon Valley in recent months, and OpenAI is at the heart of it. You may have seen DALL. E 2, a text-to-image AI software that generates images from simple written inputs. DALL.E 2 is an OpenAI-built AI system that can turn a simple text description of a gorilla dunking a basketball into photorealistic images that have never existed before.
Sam Altman and Elon Musk launched the company in 2015. Elon Musk stepped down in 2018 to focus on Tesla, but he remains a donor and advisor to the company. Their goal at the time was to advance digital intelligence in a way that would benefit humanity. At its core, OpenAI is an artificial intelligence research company that has transitioned from non-profit to capped profit. Despite no longer working day-to-day at OpenAI, Elon Musk believes that its latest program, ChatGPT, which was developed by Mira Murati, a 34-year-old Albanian woman, is frighteningly good.
But, before we get to ChatGPT, it's important to note that it's not the first attempt at an AI chatbot. Microsoft launched Tay in 2016, but within 24 hours, Twitter users branded it as rude, racist, and misogynistic, leading to its demise. Meta launched Blender Bot 3 in August however it also suffered from major flaws, providing false information on current questions at the time, such as who was the President of the USA in 2020.
OpenAI is attempting to avoid these snags by utilizing a moderation API; a moderation system that assists developers by identifying and displaying when content is generated that is not in line with their policy, such as illegal or unsafe information. It is not perfect, but it certainly helps.
ChatGPT and DALL.E 2 both use GPT3: a natural language system capable of writing, arguing, and coding. ChatGPT employs an upgraded version of this to generate texts at breakneck speeds, which brings us to ChatGPT.
ChatGPT is a computer program that can converse with humans, answer questions, and generate text based on input. It generates texts based on a prompt using deep learning, a more advanced form of AI. “We’ve trained a model called ChatGPT which interacts in a conversational way. The dialogue format makes it possible for ChatGPT to answer follow-up questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests." according to an OpenAI statement.
ChatGPT is the third update to OpenAI's generative pre-training ideas, allowing users to communicate with robots in what appears to be natural dialogue. It also corrects grammar, summarizes difficult text into simple concepts, converts movies into emojis, and fixes bugs in Python code. ChatGPT was first opened up to beta testing in November 2021 via the playground, and its release this year is expected to be an improvement over GPT3 in terms of producing fewer harmful outputs.
AI and machine learning were used to train the software with massive amounts of text data from various sources. Reinforcement learning human feedback (RLHF) was also used. ChatGPT is described as a stunningly lifelike conversation language system and the world's first truly useful chatbot in the UK's newspapers.
The first obvious effect of ChatGPT was that its own website had frozen due to high demand. With over a million people signing up to use the software in just a few days, OpenAI is forced to rapidly scale its system. Because ChatGPT is still in its early stages of development, its knowledge base is limited and mostly linked to 2021, rendering some questions obsolete.
When Elon Musk found out that OpenAI was accessing Twitter's database to train ChatGPT he directly paused access because OpenAI is no longer a non-profit and open source. There is an expectation that they should pay for this information in the future.
There are some ethical implications with this chatbot, particularly in light of recent videos in which coders use it to write code, schoolchildren use it to answer homework, and content creators use it to create content. Some troublemakers also shared how they avoided content moderation by forcing the bot to disable its ethical guidelines, generating a tutorial on how to make Molotov cocktails. Other implications include how people could be the target of generative AI. For example, deep fake videos with explicit content or propaganda questioning user privacy.
This type of generative AI may also drive people out of work. DALL.E 2 The text-to-image AI is now linked to Shutterstock, causing outrage because people are now competing with AI to sell their photos online. However, Shutterstock's CEO believes otherwise. "There are two options in this world," he said. “Be the blacksmiths who say, "Cars will get us out of horseshoes," or the technical leaders who drag people, perhaps kicking and screaming, into the new world.” However, this does not apply to photos.
Some believe ChatGPT's ability to generate human-like written text could spell the end of journalism, while others believe it lacks the nuance and critical or ethical thinking skills required. The threats that ChatGPT poses to society may appear shocking, but these threats and challenges will also help OpenAI to fix loopholes and workarounds, which is why they've released this version to the general public. They've already stated that some answers are incorrect because there is no source of truth in the data they use to train the model.
People are sometimes misled by ChatGPT's complex and detailed answers. But sometimes you have to think of the positive, like people who run online businesses who can now use ChatGPT to help customers with live chat queries that don't take all day and feel much more natural, increasing sales for small business owners.
Many in the technology industry see ChatGPT as a glimpse into the future. Some believe that generative AI is the next era-defining innovation that will change how we interact with the internet. Instead of scrolling through ads, you can ask a question and receive an accurate answer. Many users have suggested that a future iteration of ChatGPT includes a cost per question. ChatGPT will almost certainly have to pay for itself at some point; we may see advertisements in future models or there may be microtransactions.
No one can predict the future, but it could mean mass unemployment for anyone involved in the arts or it could simply be a useful tool for students to cheat on their homework. Its meteoric rise in just a few months has made it a topic of conversation, and the fact that it's already being compared to Google and the iPhone makes it sound like a football wonderkid that every manager wants to sign. OpenAI is now more of a meme machine, seeking to understand what it can and cannot do. Whatever the outcome of this new technology, we welcome you to the future.