AIGC (Artificial Intelligence Generated Content), that is, artificial intelligence content generation, is also the hottest application scenario of large models, and to a large extent, it has become a pronoun for large models. Although inaccurate, it is very representative.
Recently, after intensive contact with some clients, we have found many interesting things to share and discuss with you all.
Let's talk about some phenomena we have observed:
- In 2023, many companies were in a wait-and-see phase, and some companies started to try; in 2024, the proportion of companies waiting and seeing has decreased significantly, and the number of companies trying has increased significantly.
- A considerable number of companies are confused about whether AIGC is the goal or the means, which we can very much understand.
- The vast majority of companies only focus on content generation itself at this stage, ignoring what might be more important: judgment, decision-making, and personalization.
- Many companies have a misconception about the capabilities of AIGC, expecting too much in terms of generation, while their expectations for judgment, decision-making, and personalization are far below the actual possibilities.
- Along with the significant increase in the number of companies trying, is the number of AIGC providers. Every demand communication meeting is an opportunity to meet a new service provider. Many companies that we have heard of and have not heard of have emerged, making it very lively.
Let's talk about a little bit of thinking from MeDomino to spark discussion:
- First of all, AIGC is not the entirety of large model applications. Because the capabilities of large models are not only generation but also include judgment, decision-making, and very important personalization.
- Secondly, AIGC must be a means, not an end. To go further, AIGC should be one of the means to achieve business value, far from the only one. Even in the mobile Internet era, a certain giant (a top mobile Internet giant known for its algorithms) once said, "If you can solve problems faster with manpower, you don't need to compete on product strength." Our life science companies should still focus on business and distinguish between means and ends.
- The capabilities and boundaries of AIGC and large models are still quite limited, and everyone needs to be cautious while enthusiastically embracing them.
- AIGC should become a lever and accelerator, which means it should be based on the existing foundation, not starting from scratch.
AIGC, from capability to scenario, has not yet reached a consensus, but the path to realization is surprisingly consistent.
Nothing else, the fundamental capabilities are in the hands of large models, not relying on the engineering capabilities of companies; the scenarios are all relatively shallow, resulting in countless companies flocking to it. The core value of each participating company in AIGC has not yet been established, and homogenization competition is inevitable.
The scene of PD-1 competition will soon appear in AIGC, and will even surpass it - after all, PD-1 still requires hundreds of millions of investment and years of research and development, while the simplest AIGC only requires an account to be opened, and some companies have not even accessed the API, with costs so low that they can be ignored, which can be specifically referred to as the recent "digital people" in the technology industry.
In summary, the industry needs to rebuild consensus on AIGC and clarify some doubts:
1.Business Value - Can AIGC really improve the output of terminal content? What kind of content is suitable? What other factors are restricting it?
2.Capability Boundaries - What AIGC can and is good at doing, and what it is not good at.
3.Applicable Scenarios - What business scenarios are suitable for AIGC to solve? Besides drawing pictures, writing articles, and making PPTs, what else can AIGC do?
4.Security Compliance - Safety, compliance, and other matters that enterprise-level applications must pay attention to - in fact, this should be the first article, after all, it is the restrictive condition with the "veto power."
Finally, welcome to discuss!
