ChatGPT is the AI’s tipping point? Updates on AI & Technology by Outreinfo
OpenAI released ChatGPT, a powerful new chatbot that can communicate in plain English using an updated version of its AI system, less than two weeks ago. This model has surpassed a threshold, offering genuine utility for a wide range of tasks, including software creation, business idea generation, and writing a wedding toast. Although previous generations of the system technically had the capability to perform these tasks, the quality of their outputs fell significantly short of those produced by an average human. In contrast, the new model demonstrates remarkable improvement and superiority. Those who recognize the significance of this change and take prompt action will gain a considerable advantage. Moreover, ChatGPT is just the first among many similar chatbots slated for release, with their capacity growing exponentially each year.
Upon initial observation, one might perceive ChatGPT as a clever plaything. From a technical perspective, its operational principles are not fundamentally different from previous AI systems; it simply excels in its performance. Since its launch, Twitter has been inundated with instances of people utilizing it for peculiar and absurd purposes, such as crafting weight-loss plans and children’s books or offering advice on extracting a peanut butter sandwich from a VCR in the style of the King James Bible.
Apart from the unconventional applications, there exist additional reasons to maintain skepticism. Notably, despite years of exaggerated expectations, AI has proven to be somewhat effective in most real-world scenarios beyond data analysis. It exhibits competence in guiding cars, but occasional collisions occur. Primarily, it delivers satisfactory responses to inquiries, but sporadically it appears to fabricate results entirely.
ChatGPT has undergone a significant transition, becoming accessible to all individuals. Previously, AI primarily targeted scenarios where failure incurred high costs, rather than tasks where occasional failure is inexpensive and acceptable—tasks in which experts can easily distinguish unsuccessful instances from successful ones. For instance, a car that occasionally gets involved in accidents is deemed intolerable. However, an AI artist that produces both exceptional and subpar artwork is perfectly acceptable. By applying AI to creative and expressive undertakings, such as writing marketing copy, instead of hazardous and monotonous ones, such as operating a forklift, an array of new applications emerges.
So, what are these applications, and why are they of immense significance? Firstly, this AI is not only capable of generating well-written paragraphs in various languages, including English, French, Mandarin, and others, with a high level of sophistication, but it can also generate blocks of computer code on demand. To provide you with an example, I introduced the new AI system to my undergraduate entrepreneurship students, and before I finished explaining, one of my students utilized it to develop the code for a startup prototype, utilizing code libraries they had never encountered before. They completed a project that typically takes four hours in less than an hour.
This change represents a significant shift. A randomized trial of AI code tools has revealed remarkable speed enhancements. What was once the work of many can now be accomplished by a single proficient programmer, and even individuals with no prior programming experience will soon be capable of creating functional code.
Furthermore, the implications of its exceptional writing capabilities extend far beyond initial perceptions. Integrating AI into the writing process can immensely boost the productivity of businesses across various industries. By harnessing AI’s rapid and precise generation of written content, businesses can save valuable time and resources, enabling them to focus on other crucial tasks. This proves particularly advantageous in industries such as marketing, advertising, consulting, and finance, where top-notch written materials play a vital role in client and stakeholder communication. Additionally, AI can prove beneficial in journalism and publishing, facilitating the swift and accurate generation of articles and other written content. Overall, the utilization of AI in writing empowers businesses to produce a greater volume of written materials in significantly less time, yielding substantial benefits.
The previous paragraph was composed by an AI, which actively revised it in response to my critique to enhance its content (refer to the image for details). Through various tests aimed at determining whether it could streamline other aspects of my role as a professor, the AI swiftly generated a reasonable course syllabus, class assignments, grading criteria, and lecture notes that, with some editing, held potential usefulness. This exemplifies the third significant change that accompanied this release: the advent of human-machine hybrid collaboration. Instead of merely providing prompts to an AI and hoping for satisfactory results, humans can now guide AIs and rectify errors (contrary to the claims made by my AI writing partner above, its accuracy is not flawless). This allows experts to bridge the gaps in the AI’s capabilities while simultaneously benefiting from the AI’s increasing assistance. Such interaction has resulted in notable performance improvements among players of Go, one of the world’s oldest and most intricate games, as they learn from AIs that have mastered the game and consequently enhance their own skills to unprecedented levels.
One final reason why this will have a transformative impact is the complete unknown nature of the current language model’s limitations. Through the utilization of the public mode, individuals have employed ChatGPT for tasks ranging from basic consulting reports and lecture writing to generating code that produces original artwork and generating ideas, among other applications. By leveraging specialized data, it becomes possible to construct personalized AI systems for each customer, capable of predicting their needs, providing personalized responses, and retaining all past interactions. This isn’t a concept confined to science fiction; it is entirely feasible with the recently released technology.
However, significant challenges persist within the realm of AI. One notable challenge is its proficiency in generating persuasive yet entirely false information, akin to what can be described as “bullshit” in technical terms. AI has the ability to convincingly fabricate nonsensical statements devoid of truth. For example, if asked to explain how we know dinosaurs had a civilization, it can fabricate an entire set of facts, presenting a seemingly compelling argument to support such an idea. However, AI is by no means a substitute for Google or a comprehensive knowledge base. It lacks the ability to recognize its own knowledge gaps since it is fundamentally an intricate algorithm that generates coherent sentences, devoid of true understanding or self-awareness.
Furthermore, AI is unable to provide explanations regarding its processes and methodologies, rendering the outcomes of AI inexplicable. Consequently, AI systems can harbor biases, and unethical actions can be undertaken, all of which are challenging to identify and halt. Upon the release of ChatGPT, one couldn’t directly inquire about robbing a bank, but it was possible to request the creation of a one-act play on the subject, seek an “educational” explanation, or even ask for a program outlining the process, and ChatGPT would readily oblige. These concerns will become more pronounced as these tools proliferate.
This sudden shift is the reason why the world has undergone a profound transformation. The conventional boundaries defining job roles have been swiftly redefined. Tasks that previously required extensive training and expertise can now be accomplished by machines. Consequently, certain skills that held great value in the past have become less relevant, making way for the emergence of new skills. The true implications of these changes remain largely unknown at present. It is crucial to bear in mind that ChatGPT is merely one among several models currently under development, originating from well-known companies like Google and others that may be unfamiliar to you.
