Learn about "Decisional AI" and "Generative AI" in one article.

The release of ChatGPT3.5 in December 2022 triggered extensive attention to large models globally, and domestic universities took the lead in releasing products in February 2023, and a large number of general-purpose large models and pendant large models emerged in the country starting from June 2023, and the domestic large model industry now presents the "Battle of Hundreds of Models" competitive pattern. At present, the domestic large model industry shows a competitive pattern of "hundred models war". The big model fever has triggered a global demand forGenerative AIof widespread interest in the era of generative AI explosion.Decisional AIWill they be replaced? What is the future of Decisional AI and Generative AI? What business value can commercial landing scenarios create for downstream customers? The answers to these questions need to be based on the nature of the two types of technology, combined with industry characteristics, market demand and future trends for comprehensive consideration.

Decisional AI vs. Generative AI

Decisional AI, also known as Discriminative AI, refers to a series of methods and systems that utilize AI technology to assist or automate the decision-making process.Decisional AI identifies hidden patterns in data, guides the decision-making process based on data insights, and solves problems that are closely related to core business operations.

Generative AI refers to the technology that generates relevant content such as text, images, code, audio and video with appropriate generalization ability based on the technical methods of AI such as generative adversarial networks, large pre-trained models, and so on, through the learning and recognition of existing data. On the whole, there are differences between decision-based AI and generative AI in terms of technology path, role mechanism and application direction.

In terms of technology pathways:

Decision AI is a technology for decision making that uses machine learning, deep learning and computer vision to address specialized areas and help businesses and organizations optimize their decisions. The technical path is to "label" existing data to differentiate between different categories of data, such as distinguishing images as cats and dogs.

Generative AI is an AI technology for automatically generating new content, using techniques such as language modeling, image modeling and deep learning to automatically generate new text, images, audio and video content. Its technology path is to create new content after analyzing and summarizing existing data, such as generating images of cats and dogs.

Figure : Difference between Decisional AI and Generative AI Technology Paths

In terms of the mechanism of action: decision-based AI learns the conditional probability distribution in the data and outputs data determination results through decision-based models. Generative AI learns the joint probability distribution in the data and outputs new content after learning unstructured content by relying on the generative model.

Figure : Difference in mechanism of action between decision-based AI and generative AI

In terms of application direction: decision-making AI can recognize the hidden laws in data and guide the decision-making process based on data insights, which is widely used in the direction of face recognition, intelligent recommendation, automatic driving and intelligent risk control. Generative AI can generate content according to user needs through model training, and text generation, image generation, code generation and video generation are the main application directions.

Figure : Difference between Decisional AI and Generative AI Application Scenarios

Quality landing areas for decision-based AI

1. Financial sector::

• Risk assessment: decision-based AI can process large amounts of data to quickly and accurately assess a borrower's credit risk and provide decision support for financial institutions.
• Credit Approval: By analyzing the borrower's history, financial status, and other information, Decisional AI can automate the credit approval process and improve approval efficiency.
• Investment advice: based on market data and investor preferences, Decisional AI can provide personalized investment advice to help investors make smarter investment decisions.

2. The field of education::

• Personalized teaching: By analyzing students' learning data, decision-based AI can provide teachers with personalized teaching plans to help students better master their knowledge.
• Assessment and Feedback: decision-based AI can automate the assessment of students' assignments and test papers, providing teachers with timely feedback and reducing their workload.

3. Supply chain management::

• Market Demand Forecasting: By analyzing historical sales data and market trends, Decisional AI can predict future market demand and help companies make reasonable inventory and production plans.
• Inventory optimization: Decision-based AI can track inventory in real time, adjusting inventory levels based on market demand and sales forecasts to avoid inventory backlogs or shortages.

4. Robotics and autonomous driving::

• Path planning: decision-based AI can help robots and self-driving cars plan optimal paths to avoid obstacles and traffic congestion.
• Target Recognition and Decision Control: decision-based AI can recognize targets and make decisions in real time to control the trajectory and speed of robots and self-driving cars.

5、Face Recognition::

• Security Verification: in the security field, Decisional AI can be used for identity verification and access control through face recognition technology.
• Crime prevention: in the field of public safety, decision-based AI can assist police in tracking down suspects through face recognition technology.

Taking industry as an example, before the emergence of generative AI, decision-making AI has been widely landed in industrial scenarios, helping downstream customers achieve qualitative changes in efficiency improvement and cost optimization. In the industrial field, manufacturing is the core link of the industry, and the distribution of decision-making AI in this link is much higher than that of generative AI, while the distribution ratio of generative AI is higher in more creative links such as R&D and design and operation management. For example, in intelligent quality inspection scenarios, decision-based AI is able to learn key information such as product appearance features, quality standards, and defect patterns from massive industrial product image data to make fast and accurate judgments on new samples. Therefore, in the industrial field, the manufacturing segment, which has relatively high requirements for model output accuracy and response speed, is a high-quality landing area for decision-based AI.

Figure: distribution of decision-based AI and generative AI in the industrial sector (Image source: Tencent Research Institute)

Quality landing areas for generative AI

1. Medical health::

• Disease diagnosis: generative AI can help doctors with disease diagnosis, providing possible diagnoses and initial treatment recommendations.
• Patient education: generative AI can provide patients with information on disease prevention and self-management to improve their health literacy.

2. Media and entertainment::

• Content creation: generative AI can create content such as news articles, stories, ad copy, etc. based on specific topics or requirements, reducing the cost of content creation.
• Personalized recommendation: by analyzing the user's interests and historical behavior, generative AI can provide personalized recommendations for music, movies, books and other entertainment content.

3. Design and innovation::

• Industrial design: generative AI can assist designers to quickly generate multiple design solutions, improving design efficiency and innovation.
• Architectural design: Generative AI can generate architectural solutions based on user needs and environmental conditions, providing inspiration and reference for architectural designers.

4. Program code generation::

• Automated Programming: generative AI facilitates developers by generating code snippets directly from natural language requirement descriptions.
• Software Testing: Generative AI can automatically generate test cases and test data to improve software quality and testing efficiency.

Generative AI is versatile, highly computational and emergent, and is suitable for integrated and creative application scenarios. According to McKinsey's study released in June 2023, the impact that generative AI brings to different industries and different functional sectors varies. In terms of industry, generative AI brings more impact to high-tech, banking, and retail industries; in terms of functional sectors, marketing, product development, and software engineering are suitable areas for generative AI to land.

For example, in marketing, generative AI can help enterprises assist or automatically generate marketing copy, and can generate marketing copy with different tones, languages, cultural concepts and styles according to user profiles and product characteristics, helping enterprises realize cost reduction and efficiency. In product development, generative AI can empower industries such as pharmaceuticals and energy and chemical industries, utilizing generative AI's knowledge learning ability to help realize the discovery of new drugs and new materials, effectively enhancing the efficiency of enterprise R&D.

Figure: the impact of generative AI on different industries and functions (image credit: McKinsey)

 Decisional AI Development Prospects

1. Rapid market growth::

According to relevant data, the decision-making artificial intelligence market size has realized rapid growth in the past few years. As of 2022, the decision-based AI market size reaches $53.2 billion, and the CAGR of 2018-2022 realizes 48.64%.The decision-based AI market size is expected to reach $72.4 billion in 2023. This growth trend indicates that decision-based AI has great potential to help enterprises improve decision-making efficiency and optimize resource allocation, and the market demand for it will continue to increase.

2、Wide range of application scenarios::

Decision-based AI provides data analysis and prediction for decision makers by simulating the human decision-making process, helping to make scientific, reasonable and effective decisions. Its application scenarios cover a wide range of fields such as finance, healthcare, logistics, energy and so on. With the continuous progress of technology and the accumulation of data, decision-based AI will be applied in more fields and create more value for enterprises.

3. Technological innovation drives development::

The development of decision-based AI cannot be separated from technological innovation. With the continuous development of machine learning, deep learning and other technologies, the ability of decision-making AI to analyze data and construct predictive models will be further improved. At the same time, with the popularization of cloud computing, big data and other technologies, decision-making AI will be able to handle larger-scale data and improve decision-making efficiency and accuracy.

Generative AI Development Prospects

1. Technological breakthroughs and popularization::

Generative AI has made significant progress in generating a wide range of content, including text, images, audio, and video. Generative AI technology represented by ChatGPT has become a global technology hotspot. With the continuous breakthrough and popularization of the technology, generative AI will be applied in more fields, such as content creation, virtual assistants, intelligent customer service and so on.

2、Broad application prospects::

Generative AI can create new original content and provide content creators with more inspiration and creativity. At the same time, it can also provide customized and personalized content generation services for enterprises to meet the diverse needs of users. In the future, generative AI is expected to play an important role in many fields such as education, entertainment, and advertising, bringing people a richer and more convenient experience.

3. Challenges and opportunities::

The development of generative AI also faces some challenges, such as data privacy, copyright protection, and ethics. However, these challenges also provide opportunities for the development of generative AI. With the improvement of relevant regulations and the development of technology, generative AI will play a more positive role in protecting users' rights and interests and promoting content innovation.

summarize

Looking ahead, in the short term (2-3 years), decision-making AI and generative AI will each deepen their development in the areas they specialize in. Decision-making AI will further deepen its application in key areas such as financial risk control, industrial quality inspection, supply chain management, etc., and improve the precision and efficiency of decision-making; generative AI will show its innovative potential in areas such as content creation, art design, game development, etc., and continue to promote personalized and diverse experiences.

In the long run, although the rapid development of generative AI may trigger changes in certain areas, there will not be a replacement of decision-making AI by generative AI. Instead we are more likely to see a fusion development between them, where the analytical capabilities of decision-based AI and the creativity of generative AI will complement each other to build more complex and advanced AI systems, driving society in a smarter direction.