Factors Behind the ChatGPT Hype: What’s the Next Thing to Do?
The breakthrough in technology, a game-changer in knowledge, intelligence, and power- ChatGPT writes:
“From every corner of the globe, its voice is heard
With answers to the questions, that leave us all blurred
It’s a digital revolution, that’s changing the game
The rise of ChatGPT, it’s the new claim to fame.”
The breakthrough in natural language processing technologies signifies a major step forward in the evolution of human-machine interaction and paves the way for a more connected and collaborative world. With the advent of advanced natural language processing technologies like ChatGPT and GPT-04, the world is experiencing a new era in knowledge, communication, and human interaction and a rapid transformation in various industries, including healthcare, education, customer service, and entertainment. Likewise, machines like ChapGPT are assuming unprecedented hype and momentum for their capacity to understand and respond to human language in a mesmerizing way that offers a more efficient and personalized communication experience for individuals and businesses alike.
Let’s know about some basics in ChatGPT:
We have already heard enough about ChatGPT, which is a web-based artificial intelligence application that creates human-like text answers to queries and word prompts. You write in a query or request, and the software responds with entire, well-crafted phrases. It is powered by generative AI and designed to generate text that sounds like natural human speech, making it ideal for chatbots, virtual assistants, and other conversational interfaces.
California-based AI research and deployment company ‘OpenAI’ launched this AI-powered chatbot in November 2022. This chatbot technology is driven by a large language model (LLM) that uses natural language processing (NLP) and machine learning algorithms to understand and respond to user queries and is trained to predict the next word in a sentence after analyzing huge amounts of text. In fact, it is built on top of OpenAI’s GPT-3.5 and GPT-4 families of large language models and has been fine-tuned using both supervised and reinforcement learning techniques. GPT stands for Generative Pre-trained Transformer, which is a type of deep learning model that has been pre-trained on large amounts of text data to simulate human-like conversation, which means a GPT-driven chatbot enabled to simulate humans.
However, following its massive popularity, OpenAI releases the multimodal, bigger, and smarter GPT-4 in “ChatGPT Plus” in March 2023, which accepts both ‘text and image’ inputs and outputs text. ChatGPT Plus is the premium version of ChatGPT, which requires a subscription that guarantees subscribers access to it at the price of $20 a month. It is now banal to say that this highly anticipated and latest version of the large language model GPT-4 blows ChatGPT out of the water.
This AI bot can be used in a variety of applications, including customer service, education, copywriting, and entertainment, but typically assist users with a wide range of tasks, such as answering questions, providing recommendations, and completing simple tasks like composing emails, essays, and code. Even, this OpenAI’s white-collar worker can now browse the internet — in some instances. In addition, its first-party web-browsing plugin allows ChatGPT to draw data from around the web to answer the various questions posed to it. The plugin retrieves content from the web using the Bing search API and shows any websites it visited in crafting an answer, citing its sources in ChatGPT’s responses. But, importantly, ChatGPT has knowledge disconnection, meaning that the knowledge it’s trained on also only goes up to 2021, so it can’t work on recent material.
In essence, ChatGPT operates like a calculator. A calculator is an excellent time-saving tool. But, if you do not know what to put in it, you will not get the intended outputs. So, a group of people could use it to the best of our abilities, while others grumble about flaws and inaccuracies. Similarly, ChatGPT relies entirely on user inputs. It has endless potential if you know how to reap its benefits. Working on accurately defining what you want to achieve is the path to amazing results. To demystify, ChatGPT is still a training model, far from a complete product, as per OpenAI CEO Sam Altman’s saying, it is “incredibly limited”, and will be “a mistake to rely on it for anything important”.
The estimated monthly traffic of ChatGPT is around 96,000,000 visitors worldwide. However, ChatGPT isn’t available in Russia, China, Iran, Ukraine, Venezuela, Belarus, and Afghanistan.
A little bit on the evolutions of ChatGPT :
The journey of this giant technology is equally impressive. Around 300 billion words were fed into the system of ChatGPT, which has approximately 175 Billion parameters and is trained to understand natural language, prepared with a whopping unprecedentedly 570 GB of data, and trained with various topics and in multiple languages. The database was taken from books, encyclopedias, research articles, web texts, websites, and other types of online information, and writing was to train the ChatGPT system. It usually receives approximately 10 million daily queries.
The daily cost of running ChatGPT is estimated to be around $100,000. ChatGPT is forecasted to earn a revenue of 200 million this year and about 1 billion dollars by the end of 2024 while OpenAI has projected revenues of $200 million for 2023 and $1 billion by 2024.
ChatGPT-01 is, a sibling model to InstructGPT, trained with 7000 books. ChatGPT-02 is fueled with 40 GB of data and 8000 documents while ChatGPT-3 is trained with 45.000TB of data from internet sources. Then comes ChatGPT-04 which is 500 times more powerful than his earlier version. Simply saying, as a smart combination of artificial intelligence, natural language processing, and machine learning, GPT-4 makes things effortless and intuitive with its user-friendly design and offers an efficient, personalized experience when engaging with customer service agents that enable users to have natural conversations that feel just like interacting with an actual human being.
GPT-4 is more creative than previous language models, being able to generate, edit, and iterate creative and technical writing tasks, including writing screenplays, composing songs and even learning a user’s personal writing style. But the distinction between GPT-3.5 and GPT-4 will be “subtle” in casual conversation. OpenAI says that GPT-4 surpasses the reasoning ability of ChatGPT, and is 40% more likely to produce factual responses than GPT-3.5. However, the new model will be way more capable in terms of reliability, creativity, and even intelligence, and thus, considered as “gigantic, groundbreaking, and globally gifted”.
In fact, OpenAI’s GPT-3 large language model, is considered a foundational model for all three of OpenAI’s consumer AI apps: ChatGPT, DALL-E, and Codex. Microsoft’s Copilot is also powered by OpenAI’s GPT-4, which is more than OpenAI’s ChatGPT, embedded into Microsoft 365- Word, Excel, PowerPoint, Outlook, Teams, and more. It’s a sophisticated processing and orchestration engine working behind the scenes. As a result, OpenAI has not only attracted individual users to ChatGPT but also attracted billions in fresh investment from Microsoft. Plus, the startup has prompted an unprecedented competitive push from established tech giants like Google.
That’s why, Microsoft’s investment in OpenAI catalysts a game changer in the AI race. For example, after GPT-4 was announced by OpenAI, Microsoft has embedded ChatGPT into its Bing search engine and Azure which can interact with ChatGPT, GPT-3.5, DALL-E, and Codex. Google also announced a suite of generative AI features for Docs, Gmail, Sheets, and Slides. OpenAI’s Dall-E 2, Google’s Imagen, and start-up “Midjourney” can all generate original images whereas Meta has also announced its “Make-A-Video” tool, which turns textual prompts into short video clips. GPT-4 is also embedded into- Duolingo, the Visual assistance app Be My Eyes, and Khan Academy, as well. This way, the AI Race is getting hotter!
ChatGPT is growing faster:
OpenAI has taken the internet by storm and ChatGPT may be the fastest-growing consumer-facing App in history, as reported by multiple financial outlets. Having launched in November 2022, ChatGPT gained over a million users within a week of its launch and amassed 57 million monthly active users (MAUs) in its first month of availability. Remarkably, ChatGPT is growing faster than the social media app TikTok as it reaches 100 million users two months after launch whereas TikTok took nine months to reach 100 million MAUs, and Instagram took roughly two and a half years to hit the same benchmark. Google Translate took six and a half years to reach that milestone.
ChatGPT’s performance metrics:
No doubt, this AI technology has already revolutionized how users interact with chatbots by engaging people in conversation by recognizing the natural language and generating content quickly and easily. In fact, it has already appeared to be powerful enough to be capable of passing law and business school exams, successfully answering interview questions for software coding jobs, writing real estate listings, and developing ad content.
In the MBA program, ChatGPT did “an amazing job” and passes UPenn’s Wharton Business School test by giving answers that were correct and “excellent” in their explanations. ChatGPT performed C+ with over 95 multiple-choice questions and 12 essay questions at the University of Minnesota Law School. It also outperforms ChatGPT on human tests, including the Uniform Bar Exam (where GPT-4 ranks in the 90th percentile and ChatGPT ranks in the 10th) and the Biology Olympiad (where GPT-4 ranks in the 99th percentile and ChatGPT ranks in the 31st).
ChatGPT-4 got an almost perfect score in a Swedish University Entrance exam, Only 2.5% of the human participants get such a high score/Similar results are seen for other benchmark exams. Even, the newest version of ChatGPT passed the US medical licensing exam with flying colors. These incidents like passing a simulated bar exam with a score around the top 10% of test takers and scoring higher than 90% of SAT takers, and possessing the potential to make better clinical judgments than ‘many doctors’ tend to tell us that ChatGPT exhibits human-level performance on various professional and academic benchmarks and demonstrates the potential of chatbots like ChatGPT to assist students and professionals in various fields by providing accurate and reliable information.
ChatGPT heralds an intellectual revolution whereas GPT-4 shows “Sparks” of AGI” for it signals a true paradigm shift in the field of computer science and beyond. Moreover, GPT-5 is scheduled to complete training this December, and OpenAI expects it to achieve AGI. With a GPT-5 upgrade, generative AI may be indistinguishable from a human and will be more “exhilarating, destabilizing and transformational”. Hence, we should not onlyChatGPT, but also at AI landscape.
Usage and Impacts in real life:
AI content generated by the bot has been used in US Congress, US Congressman Jake Auchincloss read a speech also generated by ChatGPT on the floor of the House of Representatives, and also in a speech by Israel’s president, making him the first world leader publicly known to use artificial intelligence to write a speech.
In legal domains, the chatbot made its very first appearance in the recent Colombian case, where Judge Juan Manuel Padilla Garcia consulted ChatGPT about his decision. The use of Artificial intelligence (AI) for a court’s decision was made under Colombia’s new decree 2213 on AI and lawyers/judges, which may be used in civil, labor, family, administrative, and constitutional cases. Indian judges also asked OpenAI’s omnipresent tech ChatGPT to decide if a murder and assault trial defendant should be let out on bail.
In industry, Microsoft introduced Visual ChatGPT, a ChatGPT for Images that enables you to send, receive, and edit images during chatting. You can also ask questions about any images. Google has launched Bard, the search giant’s answer to OpenAI’s ChatGPT and Microsoft’s Bing Chat. Google wants Bard to become an integral part of the Google Search experience, which is still an experiment. In fact, Google wants users to think of Bard as a sidekick to Google Search, not a replacement. In February, the chatbot was shown making a factual error. Google’s value fell by $100 billion overnight. Google launches ChatGPT rival Bard in US and UK only.
China’s tech giant Baidu unveils and showcased the “Ernie” (Enhanced Representation of Knowledge Integration) chatbot, which he said could comprehend human intentions and deliver responses approaching the human level. Opera browser adds ChatGPT and AI summarization features. BloombergGPT is a 50 billion parameter language model that is trained on a wide range of financial data. GitaGTP is an AI chatbot that uses Bhagavad Gita references to answer to any questions on Krishna.
At the government level, the UK government is going to spend £900m in developing its own ‘BritGPT’ AI model. It also announced plans for regulating AI. Rather than establish new regulations, the government asked regulators in different sectors to apply existing regulations to AI. Known as the European AI Act, the rules will heavily restrict the use of AI in critical infrastructure, education, law enforcement, and the judicial system.
The U.S. hasn’t yet proposed any formal rules to bring oversight to AI technology. However, the country’s National Institute of Science and Technology put out a national framework that runs on a voluntary basis, meaning firms would face no consequences for not meeting the rules. So far, there’s been no word of any action being taken to limit ChatGPT in the U.S. Meanwhile, OpenAI faces a complaint to the Federal Trade Commission from a nonprofit research group that seeks investigation and suspension of ChatGPT releases, alleging GPT-4 is “biased, deceptive, and a risk to privacy and public safety” and violates the agency’s AI guidelines. With all of these hypes and implications, 300 million jobs are exposed globally to generative AI, such as ChatGPT. AI could eventually increase the annual global GDP by 7%. Isn’t it unprecedented!!
Whether ChatGPT should be banned or embraced:
Since ChatGPT represents a new frontier in personalized guidance and supports ‘one-size-fits-all approaches’, it would be foolish to pretend ChatGPT isn’t incredibly impressive. However, ChatGPT brings both new opportunities and new complexity.
ChatGPT is not a panacea. In the process, this AI system creates a gap between human knowledge and human understanding. ChatGPT is not without its caveats, namely its paradoxically human-seeming flaws, and biases. OpenAI CEO Sam Altman also acknowledges risks and expressed a little bit scared. The launch of ChatGPT has triggered a vigorous debate on the future of both the legal profession and legal education, as well as the evolving role of lawyers and technology in the practice of law. As of today, it still has no clear use cases in health care and legal affairs.
ChatGPT is not without its flaws. For example, ChatGPT invented a sexual harassment scandal and named a law prof as the accused, even citing a fake Washington Post article as evidence. So, attempts to use it for journalism resulted in articles riddled with inaccuracies. ChatGPT bug temporarily exposes AI chat histories to other users, which led to suspending the chat history sidebar for a period of time. In the most recent case, OpenAI confirmed a data breach due to a vulnerability in the code’s open-source library. Of late, Italy is the first country to ban ChatGPT over privacy concerns.
In particular, when it comes to higher education, it also poses to be a blend of upbeat and uneasy and forecasts the “death of the college essay era. As UC Berkeley’s experts say, “in one sense, it’s a plagiarism engine, that poses ”huge academic integrity concerns. A recent survey also suggests that most colleges, departments, and individual faculty members have yet to develop guidelines on how artificial intelligence should — or shouldn’t — be used in the classroom or research. for example, although the leading scientific Elsevier listed ChatGPT as a journal author, it later issued a correction and claims that it erroneously credited ChatGPT as an author of an academic paper.
Interestingly, the question is now not whether to use ChatGPT in schools, but how to do so safely, effectively, and appropriately. Hence, the narrative around cheating students at schools doesn’t tell the whole story. Because cheating is not a new problem: schools have already survived calculators, Google, Wikipedia, and essays-for-pay websites. In essence, ChatGPT is going to change education, not destroy it. We should underscore that generative AI could actually make learning better. Moreover, there is a handful of techniques that can detect AI-generated texts, although it has yet to attain an optimum level of performance. They include OpenAI’s classifier, DetectGPT, Turnitin’s Originality Check, AIED.UK, Copyscape, iThenticate, etc.
Another critical issue with AI language models like ChatGPT is misinformation: factually inaccurate information. Bill Gates recently published a 7-page letter about AI and his predictions for its future and expressed his concerns about AI chatbots replacing or atrophying human intelligence and the possible spread of misinformation. Crucially, ChatGPT may also hallucinate — it will generate information that doesn’t exist. These factual errors can be a huge problem. Chatbots are not getting a penalty for producing false information.
When it comes to AI biases, such a chatBot could also exacerbate existing inequalities or create new ones, particularly if access to technology is limited or unevenly distributed across different communities or groups. It also could lead to a “reliance” on technology for decision-making, potentially leading to a loss of human agency or individual autonomy.
Of late, a group of AI experts and tech executives, with more than 1,000 signatories, including Elon Musk have issued an open letter calling for a pause in the development of AI systems. with governments prepared to step in and enforce a ban. the letter has been published by the Future of Life Institute, a non-profit focused on ‘steering transformative technology towards benefiting life and away from extreme large-scale risks’.
While the open letter unlikely has little to achieve its objectives, it marks a widespread unease about AI technologies and increases pressure for more regulation. The experts also call for stronger governance systems, which should, as a minimum, include new regulatory authorities dedicated to the oversight and tracking of AI; provenance and watermarking systems to help distinguish real from synthetic and to track model leaks, and a robust auditing and certification ecosystem.
The EU’s AI ACT is setting the standard. Worldwide, standards and regulatory changes are coming, including a wave of AI standards from the International Standards Organisation (ISO), in addition to the EU AI Act and many AI Risk Assurance Frameworks. This regulation itself rests upon a risk-based analysis, with the EU grouping AI systems into four categories: Unacceptable risk: Minimal risk, Low risk, and High risk. So, the AI industry must be committed to working hard to prevent foreseeable risks before deployment.
Conclusion:
We have to move beyond thinking purely in terms of dystopianism or boosterism. Since the risks posed by AI today are more insidious, wide-ranging, and harder to foresee, such disruptive technologies must be developed and deployed in an ethical, transparent, and accountable manner. Then, we can ensure that they are used in a responsible, safe, equitable, and inclusive manner by way of taking pragmatic steps to maximize the positive impact of these technologies while minimizing the potential risks.
In addition, we should use these models as assistive tools, and constantly validate their output, as they cannot replace the value of human interaction and support. Furthermore, it should not have applications in high-stakes domains like justice systems or healthcare, unless there is some way to verify the truth. The bottom line is that the world is now experiencing a sharp shift from “knowledge is power” to “intelligence is power”. Is knowledge still power in the age of AI?