Is ChatGPT the most powerful AI product this year?Why do most people treat it like a toy, using it briefly before abandoning it? Is its functionality overhyped, or is the learning curve too steep?
High Expectations and Low Patience
Under the hype from the media and vested interests, people’s expectations for such tools far exceed what the tools can actually deliver. Often, after using them, they find that their expectations are unmet and thus give up.
Most importantly, many people still interact with ChatGPT using their search habits, unable to clearly articulate their questions, yet hoping that AI will provide satisfactory results.Although they know they need to understand their needs, learn to ask questions, and communicate effectively, they often lack the patience to slowly improve and iterate, leading to abandonment.At this point, some people are shouting that AI can help us reduce costs and improve efficiency, creating more and better content faster, signaling a transformative era; others complain that current AI is just mediocre and doesn’t effectively help them; while many are on the sidelines, still contemplating whether to invest time in exploring and experiencing it.Not Disruption but UpgradeEven though ChatGPT stands out among the new generation of AI tools, its relatively high usage threshold means that it is not the optimal choice for the average person. After a decline in traffic in June, the current growth has hit a bottleneck.However, we cannot deny the role of ChatGPT. Although it has not yet impacted the lives of most people, the foundational solutions it has explored have created new opportunities in various industries. The flourishing of domestic AI large models this year is largely related to their contributions.
Extraordinary Product Research: China’s AIGC Industry Map V3.0I hope everyone can view ChatGPT and similar tools as demo displays of the latest research achievements in the AI field.The AI products that will likely impact most people’s lives are not these single-function tools, but rather products or services that leverage their technological advancements for iterative upgrades.
Simply put, the ChatGPT that everyone experiences is just one popular product under the category of generative AI – interactive text generation. The underlying AI large model is key; with this infrastructure, more solutions and products will be created.Thus, ChatGPT may not become an essential tool for everyone,but ChatGPT + search, similar technological achievements + office tools, content generation, and various services will gradually influence the public and become part of our daily lives.For example:Search Engine + AI upgrades the search experience. The path from questioning to obtaining answers has become shorter, interactions are more convenient, and it better meets the needs of the average user.
Office Tools + AI upgrade the office experience. Our commonly used documents, spreadsheets, and PPTs have transformed into obedient assistants. You just need to describe your needs, and the rest can be completed with their help, all through dialogue.
Image Creation + AI enhances creative efficiency. Ordinary people now have the ability to create materials in different styles, simply by inputting a ‘spell’ to turn their imagined scenes into visible images, whether for new media cover images or comic materials, just a string of generation ‘spells’ can produce results.
Design teams in various industries are also widely applying AI in practice, whether for subway and bus advertisements or online posters, where we can see more and more cases of AI-assisted design creation.
Video Editing + AI lowers the creation barrier. Whether it’s video style conversion, smart editing, or material generation, for the average person, the learning cost of making a video has significantly decreased; one can go from novice to entry-level in just 1-2 weeks.
Legal + AI upgrades the consulting experience. For example, Ant Insurance launched an AI insurance planner at the beginning of the year, which not only improved users’ consulting communication efficiency but also provided ample options for users to obtain suitable insurance plans more intuitively.
Moreover, healthcare + AI, e-commerce + AI, legal + AI, customer service + AI, etc., are all future development directions for every industry and every niche. As users, we may gradually be influenced by more AI + products or services and benefit from them.So my view is: The role of AI is not to disrupt the world, but to help reshape various industries, which can also be understood as AI accelerating the iterative upgrades of various industries.From a market perspective, AI brings innovative opportunities and new possibilities; from a company perspective, AI is a solution for cost reduction and efficiency improvement, and a foundation for product and service upgrades; from an individual perspective, we gain more convenient tools that enhance our experience in certain work and life scenarios.Better search, more powerful office toolkits, better content creation tools, and more products that enhance user experience are all user needs. Therefore, the direction of AI iteration will always revolve around user experience upgrades, and this overwhelming wave will ultimately affect each of us.As for why the perception of AI’s impact differs among people, I believe it is due to the different positions people occupy. The wave of the era is still rolling in; some are still on the shore while others are in the water, hence everyone’s feelings are different.
Finding Your Own Answers
Returning to the individual perspective, the questions we may be more concerned about are: What exactly can AI help us do?
Is it how many pieces of copy can be generated in a minute using ChatGPT? Or how stunning images can be generated using Midjourney? Or using various AI tools to create works that make everyone exclaim ‘cool,’ these may not be the most important.I hope everyone can change their perspective on the issue, not focusing on a specific tool’s specific function, nor pursuing any flashy generation effects, but rather starting from your own needs, such as identifying what you want to do but cannot, what you can do but not well, and what you can do but too slowly.We can first consider: can the problems we currently face be addressed with AI tools? Which tools? How to use them? To what extent? How to replicate on a larger scale?Or simply think: Can we do what we want to do now with AI? Can it be done faster? Can it be done better?Can it be done? The answer is actually easy: it can definitely be done, but which industries and business scenarios can it be applied to? To what extent can it be done? And how can it be done? These questions still lack standard answers, which is the focus of our discussion today.My suggestion is: validate through specific practice, starting with optimizing specific workflows and comparing quantifiable results, and then address such questions.
Optimizing Workflows is Key
I believe that to leverage AI for cost reduction and efficiency improvement, the focus should actually be on optimizing workflows, especially in content production processes. Whether it’s copywriting, images, or video creation, using AI to enhance the production efficiency of certain stages will be key to achieving cost reduction and efficiency improvement through AI.AI + Copywriting WorkflowFor example, in the copywriting workflow, previously, one needed to collect materials through search engines or content platforms, then refine and organize the reference content based on their own understanding to gain inspiration and ultimately complete the creation.The addition of AI tools can help us improve the efficiency of information integration, summarization, strategy formulation, framework organization, content generation, expansion/reduction, polishing, and proofreading feedback.
Taking the creation process of a short video as an example, we can leverage AI to assist us in brainstorming and content improvement during topic planning, data association, and script and dialogue creation.
AI + Image Creation WorkflowAdditionally, for the production process of image materials, for example, creating the main visual for a cat food package, previously required purchasing copyright images or hiring cats, setting up scenes, and professional shooting to complete.
With AI tools, we can quickly generate forest backgrounds and cat images with specified actions, characteristics, and details that rival real scenes, not only at a lower cost but also meeting material effects more in line with our needs.
(Share from the NetEase Yanxuan team case)We can outline a design workflow after integrating AI, where AI can enhance creative efficiency during the phases of plan research, creative communication previews, and material generation.
AI + Video Creation WorkflowSimilarly, in the video creation workflow, there are still many parts that can be optimized using AI. In addition to the stages of confirming themes, planning scripts, and editing dialogues, we can also leverage AI to assist us in efficiently creating during material generation/matching and formal editing production stages.
From my personal experience, the efficiency of video production assisted by AI can save about half the time compared to previous workflows. I believe this same workflow optimization can also be reused in more video creation teams.
Leveraging Strengths and Abandoning Weaknesses
From the cases above, we can identify some commonalities:In the first half of the content production process, especially in the stages of logical thinking and material generation, AI’s assistance is currently significant, and the efficiency improvement is more apparent.However, during the later stages of synthesis and refinement, professional personnel are still needed for production, with AI primarily playing a supportive role at that time.
Therefore, to clarify how to leverage AI to help enterprises reduce costs and improve efficiency, we need to outline a complete workflow, clarifying what still needs to be done by humans, what can be assisted by AI, and what can be completely handed over to AI, so we can better find suitable tools, operating methods, and reference cases.For example, currently, AI can generate materials, but it cannot generate marketing posters that are aesthetically pleasing and meet requirements with one click;AI can generate the initial framework for a PPT, but it cannot create a PPT that can be presented to clients in one click;AI can generate plans and dialogues but cannot perfectly match our business scenarios and needs;AI can efficiently generate content with a score of 60, but it is challenging to produce batches of 80 or even perfect scores. We need to understand the current limits of AI to use these tools better.My advice is not to force AI to do what it is currently not good at, such as expecting AI to produce high-level, complex, and highly specialized content quickly and well, as this is currently not its strength.Answers Emerge from OptimizationSome compare the process of AI generating content to randomly drawing from a blind box, which I find quite apt. The experience and skills we accumulate in editing prompts can increase our chances of drawing rare items.However, the drawing system still has randomness; we cannot guarantee that every draw will be a success, nor can we ensure that every time we will draw something rare. More often than not, our first draw is likely to be an ordinary blind box.Similarly, to generate content that meets expectations, it usually requires multiple rounds of optimization and iteration, testing results to refine generation prompts, and then generating again, iterating multiple times until we can get close to the desired answer. This is a process that tests our professional knowledge and patience.Recently, while exploring “How AI Can Assist in Efficient Content Production for Businesses” with other business teams, I have further confirmed this point. The content generation prompts that can be practically applied in business require multiple rounds of optimization and iteration.
Regarding the practical examples demonstrated above, I expect to share a complete overview after the conclusion in October. Ah Zhao also hopes to connect with more teams’ real needs and explore specific practices of AI in different fields of content production processes together.Additionally, if you want to comprehensively learn about AIGC to enhance content marketing efficiency, and want to move from understanding to proficient use of AI tools, I recommend you take this course.👇After signing up, you can start learning. On October 19, I will create a group, and every week I will hold a live Q&A session for everyone. Any operational questions you have can be addressed.