Intuition and assumptions, common sense and logic play an important role in AI-assisted decision-making.Machine learning models can help us to better understand and analyze data by training a lot of data. We can infer unknown information based on existing data and knowledge, predict and judge results more accurately.
1与Intuition and assumptions
In the decision-making process, people often make intuitions and assumptions based on experience and feeling.Intuition refers to an intuitive perception acquired through experience and perception, which may be based on vague information or incomplete data, but can quickly produce initial judgment and decision-making on a problem.Hypothesis, on the other hand, refers to the preconditions assumed by the designer according to the user's needs and usage situation in human-computer interaction design.These assumptions are usually based on user characteristics, goals, behavior patterns, etc. to provide a more user-friendly interaction experience.In human-computer interaction design, intuition and assumptions need to be verified and corrected through user research and user testing, so as to ensure that the system is designed to meet user expectations.Through learning and model training, AI can try to simulate human intuition and hypothesis ability, which can provide intuitive reasoning and judgment in auxiliary decision-making process.
2. Common sense and logic
Common sense is the knowledge and rules people accumulate about the world in their daily lives.Logic is the rule and process of thinking, used to derive and judge.In assistant decision-making, AI can acquire and apply common sense and logic through learning and reasoning.By incorporating knowledge and rules into artificial intelligence systems, it can extract and apply common sense from a large amount of data, thereby reasoning and judging, and help people make more rational and accurate decisions.Intuition and hypothesis, common sense and logic interact in AI-assisted decision-making.Intuition and hypothesis provide the initial direction and possibility of decision-making, while common sense and logic can validate and supplement intuition and hypothesis.Artificial intelligence systems can acquire common sense and logic through learning and model training to analyze and evaluate intuition and assumptions.Combining intuition, assumptions, common sense, and logic, AI systems can provide comprehensive decision support and help people make smarter and more reliable decisions.
3常识Anti-common sense and anti-logic
In artificial intelligence-assisted decision-making, there are sometimes anti-common sense and anti-logical situations.This is mainly because AI systems may rely on specific data and algorithms in making decisions regardless of human common sense and logical reasoning.A common counter-common sense is that artificial intelligence systems may draw false conclusions based on statistical data.For example, in predicting the survival rate of patients with disease, the system may simply predict on the basis of relevant factors in historical data, ignoring the patient's specific situation.This may lead the system to draw conclusions that are inconsistent with the actual situation, as individual differences and special circumstances are not taken into account.Another anti-common sense is the data bias that can occur in artificial intelligence.Training data of artificial intelligence systems are often based on historical data, which may be biased.If the system relies solely on these data for decision-making, it may be prejudiced or discriminatory against certain groups or social issues.For example, a facial recognition system may be more likely to misidentify the faces of people of African descent, since the system's training data may contain only a small number of samples of African descent.In addition, anti-logical situations may occur in AI systems.This is because artificial intelligence systems are often trained and decision-making based on large amounts of data and complex algorithms that do not necessarily follow human logical thinking.For example, in natural language processing tasks, artificial intelligence systems may produce answers that seem plausible but are not common sense.This is because the system simply generates answers based on data and models rather than logical reasoning or common sense judgment.Here are some other AI-assisted decision-making systems that may produce anti-common sense and anti-logical results in some cases:
a.During the loan approval process, the traditional loan approval model may consider the credit score, income level and other factors to make the decision.However, AI-assisted decision-making may introduce non-traditional factors such as social media activities, shopping behavior, etc. to assess the applicant's credit risk.This anti-common sense may lead to qualified applicants being refused loans because they share some bad behavior on social media, but it doesn't mean they won't be able to repay the loan on time.
b. In medical diagnosis, AI-assisted decision-making may give an incomprehensible or counterintuitive diagnosis based on big data analysis results.For example, a patient's symptoms do not match a known disease, yet a big data-based model gives an uncommon but highly probable diagnosis.It may be difficult for doctors to accept this result because it contradicts empirical knowledge and produces anti-common sense decisions.
c. In self-driving cars, AI-assisted decision-making may choose a seemingly unreasonable action in an emergency.For example, AI may choose to hit pedestrians on the side of the road, rather than normal avoidance, when avoiding crashes.This counter-logical decision may be due to AI's finding other potentially more serious consequences in risk assessment, preferring lesser harm.
In these cases, AI-assisted decision-making may break through the boundaries of human common sense and logic and use a lot of data and model analysis to make decisions.However, such anti-common sense and anti-logical decision-making results need to be thoroughly verified and reviewed to ensure their accuracy and rationality.
In order to solve these problems, the design and training of artificial intelligence system need to take into account human common sense and logical reasoning.This can be achieved by introducing more human intervention and oversight to ensure that systematic decisions are consistent with common sense and logic.In addition, bias and discrimination could be reduced through more balanced and diversified training data in order to improve the quality and fairness of systematic decision-making.