Complex AI: Key ingredients for enterprise AI success

According to a recent global report released by Dynatrace, driven by increasing investment in artificial intelligence, "composite artificial intelligence" will become a key factor for enterprises to successfully apply artificial intelligence. Dynatrace pointed out that while 83% of technology leaders believe that artificial intelligence is essential, 95% believe that generative artificial intelligence (GenAI) will be more effective if it is assisted by other types of artificial intelligence

The report points out that enterprises need to adopt a comprehensive approach to artificial intelligence that combines multiple types of artificial intelligence. These types may include GenAI, predictive and/or causal AI, and data from disparate sources such as observability, security, and business events. Through this approach, enterprises can achieve higher levels of reasoning and make the output of artificial intelligence more accurate, relevant and meaningful

It can be seen from this that artificial intelligence is not a a simple technique. It involves a variety of types, methods, data, and scenarios, each with its strengths and limitations. To realize the full potential of artificial intelligence, a single artificial intelligence technology is not enough. There is a need to combine different AI technologies and data sources to provide more advanced inferences and more accurate, meaningful and contextual AI outputs. This is the concept of composite artificial intelligence. It refers to combining multiple types of AI, such as generative, predictive, and causal, with different data sources, such as observability, security, and business events. Composite AI is a key factor in successful enterprise adoption of AI. Because it can help enterprises solve complex problems, provide more valuable solutions, and better adapt to changing environments

This article is based on a global report by Dynatrace to analyze compound The concepts, applications and value of artificial intelligence, as well as the advantages and challenges of composite artificial intelligence. Our purpose is to better understand the importance and potential of compound artificial intelligence, and how to effectively adopt compound artificial intelligence to enhance the competitiveness and innovation of enterprises

01 Compound Artificial Intelligence The concept of intelligence

As we all know, artificial intelligence is an interdisciplinary subject involving computer science, mathematics, statistics, psychology, philosophy and other disciplines. It enables computers or machines to possess human-like intelligence, such as perception, understanding, learning, reasoning, decision-making, creation and other abilities. The research and development of artificial intelligence has a history of several decades, during which it has experienced many ups and downs and changes. This has led to the formation of multiple types, methods, data and scenarios of artificial intelligence technology, each with its own advantages and limitations

The recent rapid development of generative artificial intelligence Intelligence (generative AI) uses deep learning and other technologies to generate new content from data, such as text, images, audio, etc. The advantage of generative artificial intelligence is that it can help us create new products and services, such as virtual assistants, personalized recommendations, intelligent content, etc. However, generative artificial intelligence also has some limitations and may be inaccurate, unreasonable or unethical. For example, the generated content is inconsistent with the facts, violates common sense or hurts the feelings of others.

The history of the development of predictive artificial intelligence (predictive AI) can be traced back to the 1950s, when the first batch of statistical learning theories and machine learning algorithms appeared. Predictive artificial intelligence uses statistical learning and other technologies to discover patterns and trends from data for classification, regression, clustering, etc. The advantage of predictive artificial intelligence is that it can help us optimize existing products and services, such as performance monitoring, fault prediction, anomaly detection, etc. The limitations of predictive artificial intelligence are that the predicted results are inconsistent with reality, are affected by noise or anomalies, or lack explainability.

Causal AI is a kind of utilization Artificial intelligence that uses data and algorithms to infer cause and effect. It helps organizations understand the causes and effects behind data, leading to better decisions and interventions. Some application fields of causal artificial intelligence include medicine, social sciences, economics, education, policy making, etc. The core concept of causal artificial intelligence is the causal diagram, which is a graphical model that uses nodes and arrows to represent variables and causal relationships. Cause-and-effect diagrams can be used to represent the mechanisms by which data are generated and how the distribution of data can be changed through intervention or experimentation. Cause-and-effect diagrams can also be used to answer causal questions, such as "What happens if I do this?" or "Why does this variable affect that variable?"

Rewritten in Chinese as follows: In addition to different types of artificial intelligence technologies, there are also different data sources available, including observability, security, and business events. Observability data refers to data related to application performance, software development and security practices, IT infrastructure, and user experience. Security data refers to data related to cybersecurity, data protection and privacy compliance. Business event data refers to data related to business processes, transactions, customer behavior and satisfaction. Different data sources provide different information and value, but also face different quality, security and privacy requirements and challenges

Compound artificial intelligence is the combination of different types of artificial intelligence technology and data Sources are combined to provide more advanced reasoning and more accurate, meaningful and contextual AI output. Compound AI can help us solve complex problems, provide more valuable solutions, and be better able to adapt to changing environments. The core of composite artificial intelligence is to use the advantages of different artificial intelligence technologies and data sources to make up for their respective limitations and achieve synergy and complementarity between artificial intelligence technologies and data sources

02 Integration and adjustment of compound artificial intelligence

Integration and adjustment of compound artificial intelligence is the combination of different types of artificial intelligence technology and data sources to provide more advanced reasoning, and more accuracy , more meaningful and contextual AI output. The integration and adjustment of composite artificial intelligence requires consideration of multiple aspects, such as the goals, scope, methods of artificial intelligence, the quality, security and privacy of data, and the evaluation and supervision of artificial intelligence. Therefore, some basic concepts and principles are needed

The goal of artificial intelligence is to solve problems or implement functions, such as diagnosis, prediction, generation, recommendation, etc. These goals determine the type, format, and content of AI output, as well as its value and impact. In order to facilitate design, implementation and evaluation, the goals of artificial intelligence should be clear, specific, quantifiable and achievable

When determining the scope of artificial intelligence, the areas involved need to be considered or Scenarios, such as medical, education, finance, manufacturing, etc. The scope of artificial intelligence determines the sources, characteristics and requirements of its input and output, as well as the complexity and difficulty of the input and output. In order to facilitate the acquisition, processing and application of artificial intelligence, the scope should be appropriate, reasonable, feasible and scalable

Selection methods for artificial intelligence include the use of generative, predictive and causal artificial intelligence Intelligent technologies or algorithms, as well as deep learning, statistical learning, causal reasoning, etc. These methods determine the quality, reliability, and interpretability of AI output, as well as the strengths and limitations of AI output. Artificial intelligence methods should be suitable, effective, innovative and improveable in order to be optimized, verified and updated

Data quality refers to the accuracy, completeness and consistency of data , timeliness and relevance. The quality of data directly affects the accuracy, rationality and ethics of artificial intelligence output, as well as the trust and acceptability of artificial intelligence output. Therefore, the quality of data should be high, stable, clear and new to facilitate the learning, reasoning and generation of artificial intelligence.

The content that needs to be rewritten is: data security also needs to be considered and privacy protection, including data encryption, authorization, auditing, backup and deletion, etc. The security and privacy of data have an impact on the security, legality and accountability of AI output, as well as the risks and consequences. Therefore, data security and privacy should be strong and strictly adhere to regulations and laws to facilitate the prevention, response and correction of artificial intelligence.

is very important for the evaluation and supervision of artificial intelligence , its output needs to be inspected and improved. This includes work on testing, evaluation, feedback and monitoring of AI output. The evaluation and supervision of artificial intelligence have implications for the efficiency and effectiveness of its output, as well as innovation and development. Therefore, the evaluation and supervision of artificial intelligence should be continuous, comprehensive, objective and timely in order to facilitate its adjustment, optimization and update

The integration and adjustment of composite artificial intelligence is a process that involves A multi-faceted process designed to deliver more advanced reasoning capabilities and more accurate, meaningful and contextual AI output. This is a process that requires a significant investment of time, resources and expertise, as well as a high degree of coordination and cooperation

03 The application of composite artificial intelligence

The application of composite artificial intelligence is to use different types of artificial intelligence technology and data sources to solve complex problems and provide more advanced reasoning and more valuable output. This approach combines multiple types of artificial intelligence to achieve more advanced reasoning and better performance. The application fields of compound artificial intelligence are very wide. For example, in the field of natural language processing, compound artificial intelligence can be used to generate, understand and translate natural language, including the use of generative AI, predictive AI and causal AI and other technologies

The application of compound artificial intelligence can be applied to various fields and industries such as medical care, education, finance, manufacturing, entertainment and security

In the medical field, compound artificial intelligence Generative, predictive, and causal AI can be combined with observability, security, and business event data to deliver more accurate diagnosis, treatment, and prevention options, as well as more personalized care. For example, generative artificial intelligence can generate new medical information, such as lesions, mutations, diagnoses, etc., through data such as medical images, genetic sequences, and medical records. Predictive artificial intelligence can discover patterns and trends from medical monitoring, testing, evaluation and other data, such as condition, risk, effect, etc. Causal artificial intelligence can infer causal relationships from data on medical intervention, treatment, prevention, etc., such as drugs, surgeries, lifestyle, etc. Observability data enables monitoring and analysis of medical devices, systems and processes, improving availability and reliability. Security data can protect and comply with medical data and privacy, preventing data misuse or leakage. Business event data can measure and improve medical services and satisfaction, increase revenue and profits

The application of compound artificial intelligence in the education field is to combine generative, predictive and causal artificial intelligence and learner behavior, ability and preference data combine to provide us with better teaching, assessment and feedback, and learning paths that are more suitable for us. For example, generative artificial intelligence can create new teaching content for us, such as questions, answers, examples, etc., based on data such as textbooks, courses, and knowledge maps. Predictive artificial intelligence can help us discover rules and trends such as learning level, progress speed, and learning style based on our learning activities, results, feedback and other data. Causal artificial intelligence can infer the causal relationship between teaching methods, learning resources, learning outcomes, etc. based on our teaching interventions, strategies, effects and other data. Learner behavior, ability and preference data can allow artificial intelligence to understand us better, provide us with personalized and customized services, and make our learning more efficient and effective

Compound Artificial intelligence plays an important role in the financial sector. It uses generative, predictive and causal artificial intelligence to combine market, customer and risk data to provide us with better financial solutions and services. For example, generative AI can provide us with new financial information based on various data, such as quotes, recommendations, strategies, etc. Predictive artificial intelligence can help us discover financial rules and trends based on various data, such as prices, demand, behavior, etc. Causal artificial intelligence can help us analyze financial causal relationships based on various data, such as impact, risk, return, etc. Market data can help us better understand and analyze market conditions and improve market transparency and efficiency. Customer data can help us better understand and serve customers, and improve customer loyalty and satisfaction. Risk data can help us better identify and manage risks, and improve the controllability and predictability of risks

04 The value of composite artificial intelligence

The rewritten content is: The application of composite artificial intelligence can improve the reliability and applicability of artificial intelligence. By combining different types of artificial intelligence technologies and data sources, composite artificial intelligence can make full use of the advantages of various artificial intelligence technologies and data sources to make up for their respective shortcomings and achieve collaborative work and complementary effects. In this way, composite AI can provide more comprehensive, accurate, reasonable and ethical AI output to support key business scenarios and avoid inconsistent, inappropriate or undesirable issues

Bernd Greifeneder, CTO of Dynatrace, said: “Getting GenAI to generate meaningful content that users can trust to solve a specific problem is a difficult task. The team needs to invest a lot of time and effort in designing suitable prompts that GenAI understands what should be done. What kind of content is generated and verifies the accuracy of the generated content.”

To achieve this, he emphasized the importance of understanding that different artificial intelligences are not one-size-fits-all. Many application scenarios of GenAI, especially complex scenarios such as optimizing software code or solving security vulnerabilities, require the use of composite artificial intelligence (composite AI). Composite AI refers to the combination of different types of AI, such as causal reasoning AI (which can find out the causes and effects of system behavior) and predictive AI (which can predict future events based on past data), providing the necessary Background information

Enterprises that develop the right strategy and combine various artificial intelligence technologies with high-quality data can significantly improve the efficiency of development, operations, and security teams. , while delivering lasting business value

Compound AI is an approach that combines different types of AI technologies and data sources to provide more advanced reasoning and more valuable output. We can combine generative AI (generating new content from data), predictive AI (predicting future events based on past data), and causal AI (identifying cause-and-effect relationships in data) to generate more accurate, more informative meaning and more contextual content. Additionally, we can leverage different data sources such as observability data (used to monitor and analyze system performance and status), security data (used to protect and comply with data usage and transmission), and business event data (used to (for measuring and improving business processes and effects) to ensure that artificial intelligence can use more comprehensive, reliable and relevant data

Reference: https://aimagazine.com/data -and-analytics/composite-ai-could-drive-success-as-investments-increase Reference link mentions that as investment increases, compound AI may drive success

The above is the detailed content of Complex AI: Key ingredients for enterprise AI success. For more information, please follow other related articles on the PHP Chinese website!