AI is like a new continent, full of great wealth and endless opportunities waiting for us to explore. However, the mining of these wealth and opportunities is not achieved overnight, and needs to be based on the long-term efforts and continuous attempts of the predecessors. AI still has a long way to go, it is just a beautiful vision at present.

In fact, AI does not have enough data support at this stage, and many startups exaggerate the extent to which artificial intelligence is used in business operations. Although artificial intelligence and machine learning are hailed as revolutionary technologies that are expected to create huge value for the company, we cannot ignore the difficulties and pain points that AI technology faces in the process of product landing.

The industry lacks uniform standards

Like the rise of mobile applications, the AI ​​industry is in urgent need of a new cross-platform system, just as Android and iOS have laid the cornerstone for mobile applications. This system will cover everything from models and algorithms to software implementation and operating system. It is characterized by multi-data source compatibility (such as multiple databases), multi-programming language compatibility (such as TensorFlow, Python), and the ability to run seamlessly on various hardware devices to realize the dream of cross-platform AI.

However, the current AI industry is facing a common problem, that is, the lack of unified standards. Without unified standards, large-scale production and manufacturing cannot be realized, and there is still a long way to go for product technical form display and business model operation.

Expensive

With the advancement of artificial intelligence technology, more and more emerging companies have released their own intelligent brains and provided them to users through cloud services. After these companies have attracted a large number of users, they are facing a trouble: the running cost of the server is too high, and some companies have not considered this problem due to lack of funds and technical strength.

The training of artificial intelligence models can be described as a huge challenge, requiring a lot of resources, time and money. Some cloud service providers provide automated model training and usage services, but the cost of these services is quite expensive, especially in terms of operating costs for intelligent brains. If the service fees provided by start-up companies cannot cover the operating costs of servers, the survival and profitability of these companies will be greatly affected, and naturally they will not be able to commercialize on a large scale.

Concentrated AI business implementation, serious losses for enterprises

The security and financial industries are the most widely used fields of artificial intelligence in the real economy, accounting for the vast majority of the market share, reaching about 70%. However, this also means that in other industries, there is still a lot of room for development in the deep integration and implementation of artificial intelligence.

Even though some unicorn companies have made some progress in the implementation of applications, most of them are still facing losses. Although the prospect of AI is good, not every enterprise can survive in this track.

Legal and ethical issues facing AI

With the widespread application of artificial intelligence, there are some moral and legal risks, such as: privacy and data abuse, bias and discrimination, transparency of autonomous decision-making, addiction and abuse of AI, and liability and legal disputes.

In order to deal with the moral and legal issues arising from new technologies, countries have also introduced laws and regulations one after another.

The promulgation of laws provides institutional guarantees for the implementation of AI scenarios, but it also takes a certain period of transition from the law to implementation. During this period, many commercial AI scenarios still face institutional defects. This has also led to one of the reasons why it cannot be commercialized on a large scale in the short term.

Rapid development period, bubble period, rational development period, mature period and other stages. In this process, many companies have entered the stage of AI one after another. However, in the process of industry competition and technological iteration, after the bubble, when the real competition of technical strength and overall comprehensive capabilities, a large number of companies will slowly die again. Only when costs are reduced, various systems and laws are mature, and AI returns to the normal development state of the industry, can commercialization scenarios be truly realized.