Artificial Intelligence technology is the development of the ability of computer systems to carry out tasks involving thought processes including reasoning, planning, learning and self-correction.
The term Artificial Intelligence was first introduced in 1956 by John McCarthy, a computer scientist and currently Artificial Intelligence technology is one of the triggers for the industrial revolution 4 so that getting to know Artificial Intelligence technology more closely can prepare the best strategy for the next achievements in the future.
Difference between Artificial Intelligence Technology and Human Intelligence
Artificial Intelligence Technology:
- More objective because without involving emotions.
- Results are accurate but allow for error.
- Adaptability takes a long time.
- Not able to multitask.
- No social interaction
- Optimization Function
- digital phase
- This intelligence has existed since ancient times.
- Slow to do certain tasks.
- Decision making tends to be subjective because it involves emotions and social interactions.
- Results are still less accurate.
- Quick and easy adaptability
- Can multitask.
- Doing social interaction
- Innovation Function
- Analog phase
Artificial Intelligence Technology Grouping
1. By task
Artificial Intelligence technology based on its duties is divided into 2 categories, namely AGI (Artificial General Intelligence) and ANI (Artificial Narrow Intelligence). AGI is Artificial Intelligence in general tasks and is still hypothetical as seen in science fiction films while ANI is Artificial Intelligence in specific tasks only.
2. Based on the way of work
Artificial Intelligence technology based on how it works is divided into 2 categories, namely rule based Artificial Intelligence and machine learning.
- Rule based Artificial Intelligence works based on human programs to map computer output when there is input information to the system.
- Machine learning does not rely on the rules generated by human programs but performs data processing and looks for certain patterns in the data.
Artificial Intelligence technology based on its use is divided into 4 categories, namely Natural Language Processing (NLP), speech recognition, robotics and computer vision.
- Natural Language Processing and speech recognition are able to imitate human speech and language abilities, for example Siri, Alexa, Google Assistant.
- Robotics creates machines to perform various physical functions in place of human skills.
- Computer vision is a branch of Artificial Intelligence technology in processing images, photos, videos and direct camera photos. Examples of the application of Artificial Intelligence computer vision are Facial Recognition and CCTV detection to see possible traffic violations resulting in electronic fines.
The category of Artificial Intelligence technology based on how it works is the most popular today is machine learning because it can solve problems without being programmed based on large amounts of data today and more in the future. Data is the fuel for Artificial Intelligence growth, the more data the greater the growth of Artificial Intelligence.
Machine learning is divided into 2 categories, namely Neural Network and Deep Learning. Neural Network is a machine learning technique that recognizes data patterns by imitating the way the human brain works, while Deep Learning is a variation of Neural Network implementation using several hidden layers to extract data features without human assistance.
How can machine learning learn from data and derive its intelligence?
Machine learning is a computer program that has the ability to learn from an experience (E) if the performance (P) of doing a certain task (T) increases because of that experience (E).
To create an Artificial Intelligence project, one of them can take advantage of the Artificial Intelligence Canvas framework. Artificial Intelligence Canvas is a format for analyzing the potential for applying Artificial Intelligence technology in work processes or business processes of companies and organizations.
The way to use Artificial Intelligence Canvas is to fill in the answers and instructions for each of the important elements of the Artificial Intelligence Canvas.
- What tasks/decisions are examined? (Briefly describe the task being analyzed)
- Predictions: identify the main uncertainties!
- Considerations: set the outcome to be right versus to be wrong!
- Actions: what actions to choose from?
- Results: select a performance measure to assess whether it has achieved results!
- Input Data: What data to generate predictions once owning the Artificial Intelligence algorithm?
- Data Training: input data and past actions and results to train Artificial Intelligence and make better predictions?
- How will this Artificial Intelligence impact the entire workflow? Describe here how the Artificial Intelligence for this task/decision will impact the associated tasks in the overall workflow. Will it cause staff turnover? Will this involve staff retraining or job redesign?
Artificial Intelligence development process sequentially is start, problem definition, data compilation and data collection, data understanding, data preparation (data cleaning), developing prediction model, parameter tuning (model tweaking), model evaluation, deployment, feedback and finish.
Problem Definition
The definition of this problem requires an in-depth understanding of the background conditions of the organization or business.
Data Compilation
After the problem is clearly defined, then prepare the data requirements at the data compilation stage in the form of content and format.
Data Collections
Start seeing if the data actually exists and can be accessed. If it turns out that the data is not available and obtainable, it may be necessary to revise the data requirements or indeed have not been able to make the Artificial Intelligence system according to the problem, so you have to think about how to collect this data requirement.
Understanding data
Start looking for insights and insights using descriptive statistics and data visualization, usually at this stage you will gain insight into the completeness or quality of the data.
Data Cleaning
Armed with the insights from the previous stage, do data preparation or data cleaning because often the data is still dirty, untidy or incomplete.
Predictive Models and Parameter Tuning
Choose the machine learning technique that best fits your data and task.
Model Evaluation
This stage aims to ensure knowing the quality of the model.
Deployments and feedback
Install the model in a production environment so that it can be used immediately.
Metrics that measure the quality of Artificial Intelligence in generating predictions:
- Precision
- recall
- accuracy
- F1 Score
Recall is a measure of correctly identified positive results of all positive results.
Accuracy is the result of correctly identified predictions.
F1 score is the average of harmonic precision and recall data.
Data and Tools
The most important component of Artificial Intelligence technology is data. Data is a collection of descriptions of an object and event.
Data is divided into 2 types, namely structured data and unstructured data.
- Examples of structured data: data tables and databases.
- Examples of unstructured data: text, video, images and all of them don't have an identity (label).
The more data, the more patterns can be recognized by the Artificial Intelligence system so that the ability of Artificial Intelligence to predict something is more accurate.
Important Artificial Intelligence development tools
- Programming Language: Python
- Data Collection: Mongo, MySQL, PostgreSQL
- Data Preparation: Pandas
- Modeling Machine Learning: scikit learn.
- Modeling-Deep-Learning: Keres, TensorFlow, Py Torch
- Deployment: amazon web service or google cloud platform
As an example of the use of Artificial Intelligence technology, namely in the field of customer service. There are 3 NLP based solutions in this area which are:
- Artificial Intelligence Chatbot: Automating the process of answering customer questions through text message channels such as WhatsApp and being programmed to have the most conversational threads asked by customers so that the majority of customer questions can be answered without human assistance.
- The NLP Model Assist Agent will read customer messages before they are read by cs agents and generate message response program recommendations.
- Artificial Intelligence Voice utilizes speech recognition technology to translate customer telephone speech into text which is then processed by an NLP model such as a chatbot to detect the meaning and intent of the speech so that it can answer consumer questions.
1. Google Assistant (Natural Language Processing)
Designed to carry out a series of specific tasks such as sending messages, playing songs, opening certain applications and others on smartphones using speech recognition so that they are able to translate the natural pronunciation of people's languages into computer instructions.
2. Google Maps
Google Map provides route information and recommendations as well as update information by collecting GPS data for all application users and then detecting the position and speed of millions of vehicles around the world. Artificial Intelligence technology can predict traffic density and when entering a destination location, it will calculate the distance and travel time of all travel routes and finally choose the route with the shortest distance and travel time.
Artificial Intelligence technology from an economic and business point of view
From an economic standpoint, Artificial Intelligence technology will make the prediction process cheaper. This is because today's Artificial Intelligence technology is basically machines that effectively predict things.
The impact of Artificial Intelligence technology on work
- Artificial Intelligence makes work easier.
- Artificial Intelligence reduces the number of certain jobs.
- Artificial Intelligence is changing work processes.
- Artificial Intelligence changes human abilities in certain jobs
Artificial Intelligence technology actually drives prospects for economic growth and jobs along with overhauling infrastructure, industry and the economy. For example, in the development of machine learning, it still requires a lot of human roles to ensure that the data needed by machine learning is of high quality and does not cause bias.
On the other hand, Artificial Intelligence technology provides business and economic benefits such as increasing productivity and innovation in the industry. The impact on jobs varies, such as some jobs disappear but new jobs appear.
Loading posts