AI / Machine Learning

From New Media Business Blog

Jump to: navigation, search

Contents

AI Machine Learning

Machine learning is a subfield of artificial intelligence that utilizes data and algorithms to mimic human cognitive behavior [1].

Automated warehouse

Automated warehouse picking is the process of automating the movement of inventory into, within, and out of warehouses to customers with minimal human assistance. It works by using software and technology like robotics and sensors to automate tasks. Automated warehouse picking is more suited for large volume warehouses and distribution centers with enough space to accommodate the specialized equipment. [2]

How does an automated warehouse work?

An automated warehouse helps ensure that customer demand is met with the business-critical operations in the company. It works by starting with WMS aka warehouse management system that automates manual processes and data capture, inventory control, and data analysis. Then the systems integrate with other solutions to efficiently manage and automate tasks across different business practices and supply chain functions. [3]

Amazon robotics

Amazon robotics
Amazon robotics help develops sophisticated machinery and software to optimize efficiency in Amazon fulfillment centers. It uses its software and machinery to automate the flow of inventory which include 3 main physical components to the system, which are mobile shelving units, robots, and employee workstations. Amazon robotics is expected to build models in the future that can further automate package tracking and help automate package damage assessment. [4]

Pros of automated warehouse

Increase productivity

Increased productivity will no doubt be the first thing that is noticed when the automated warehouse is used, whether a company selects an automated retrieval storage system, a fully-autonomous picking vehicle, or a lightweight collaborative system. [5]

Keeping the existing infrastructure

When the automated picking system first hit the market, it featured clucky elements that often required bolts, lots of available space, and sometimes huge changes in infrastructure. However, these days automated warehouse picking is lightweight and can be programmed to exist or move within the warehouse. It helps avoid the need for costly overhauls and possible compliance problems. [6]

Sync with software for on-demand reporting

It allows for valuable information, such as inventory counts, to be generated automatically and accessed anytime. [7]

Cons of automated warehouse

Expensive initial investment

A company can easily spend upwards of a million dollars or more just on the initial investment alone, depending on the picking system it chooses. It is not to be taken lightly, especially if the company is still in the beginning stage of defining its long-term goals, budget, and leadership. [8]

Lack of adaptability

It is no doubt that automated warehouse picking surpasses human beings working in the same roles. However, these systems have yet to reach a point where adaptive thinking can take place. For example, the automation must be reprogrammed by humans and technical errors must be fixed manually. [9]

Technological unemployment

One of the most concerning problems is that the human workforce will inevitably be replaced by technology. Even though technology helps create a surge in productivity, this can cause a massive rise in unemployment. In addition, a company’s team morale might hugely decrease due to the fact that employees might feel that their jobs are in danger and they might get fired. [10]

Speech Recognition

Speech recognition is technology that utilises artificial intelligence (AI) to record, analyse, and respond to human voices. The process involves human speech being converted to text, filtered from noise, and analysed using neural networks and algorithms, which function similar to human brains, where each word is mapped to some definition and characteristics to help in putting everything together to derive the intended outcome. After this process is complete, the right response is sent back to the user.

Here is how the algorithm for speech recognition technology works[11]. First, humans speak to their smart devices to record their speech, which then is converted to text and is filtered out of any noise that was also recorded. Then, the information goes through neural networks to map what the user said to what is available in the existing pool of data and the technology tries to understand the context using artificial intelligence. Once it has understood what the user is looking for, it constructs and evaluates a response it deems to be correct before sending it back to the user.

Training Speech Recognition

The more speech recognition is trained, the more accurate, personalised, and conversational the responses will become. Training involves understanding each human’s unique voice characteristics such as tone, pitch, dialect, and pace[12], background noise generated from their environments, and the complexity and frequent use of words. As a result of this training, the results generated by speech recognition create a more human-like communication service that understands the user very well and what they are trying to achieve. Therefore, the user will have a more personal and quicker experience when using speech recognition to complete their tasks.

Speech Recognition Market

The speech recognition market is expected to reach approximately $27 billion by 2026[13] and $45 billion by 2032[14], led by companies such as Amazon, Alphabet, Apple, and Microsoft. Research continues to be conducted and more advanced and thorough algorithms are worked on to improve speech recognition, with the goal of eventually replacing specific human jobs and tasks, especially customer service.

Applications of Speech Recognition - Consumers

Digital Assistants

Apple Siri Digital Assistant
Consumers are using speech recognition technology through their smart devices, including phones, watches, speakers, TVs, and vehicles. Virtual assistants such as Apple Siri, Microsoft Cortana, Amazon Alexa, and Google Assistant[15] help automate various tasks that can be tedious at times, like searching or shopping online, setting reminders, or answering messages and phone calls.

Integrations with Digital Assistants

Amazon Echo Dot
In addition, companies are utilising speech recognition technology to give consumers a more seamless experience with their products and services. PayPal[16] allows users to make online payments through voice commands using Siri, and Starbucks[17] allows users to automatically place orders to the nearest Starbucks using Alexa. These quality of life additions to consumers’ lives help eliminate wasting time on smaller tasks and provide more options for consumers depending on their needs.

Applications of Speech Recognition - Businesses

McDonald’s Automated Drive-Thru

McDonald's Automated Drive-Thru
McDonald’s and IBM entered a strategic partnership to help automate drive-thru lanes at McDonald’s fast food chain restaurants[18] to speed up orders and allow employees to focus on more value-added activities such as making, preparing, and packaging higher quality meals for its customers. Customers are able to speak to a kiosk that uses IBM’s Apprente speech recognition technology to try to understand and process orders. Although the tests have been well-received, the speech recognition is able to understand about 85% of orders, with employees having to step in from time to time[19] if confusion or misunderstanding arises during the ordering process.

Hyundai’s Dynamic Voice Recognition System

Hyundai Dynamic Voice Recognition System
Hyundai partnered with SoundHound to allow users to obtain information and purchase almost anything using Houndify speech recognition technology built into their Elantra vehicles. This opens up great potential as drivers would be able to shop the latest deals on products, book reservations at restaurants and hotels, and more, providing users the ultimate experience. As stated on Hyundai’s website, “Powered by Houndify, the Dynamic Voice Recognition system makes the driving experience safer, easier, and more convenient by providing drivers with access to cloud-based information from hundreds of knowledge domains, all hands-free.”[20]

Voice Biometrics

Banking, insurance, health, and other industries that require high levels of security and identification processes can take advantage of each human’s unique voice characteristics to use as a biometric to better thwart cyberattacks and other dangerous hackers seeking to leak or acquire personal data. Voice biometrics use 100+ voice characteristics (including the ones covered in this blog) to authenticate users[21], and have proven to be a reliable method of identification. For example, Bank of America[22] is investing in this technology to provide better customer experiences, reduce friction in the identification process, and ensure trust and safety of their users’ data. Also, hospital patients are beginning to use their voices to access their Electronic Health Records[23] more quickly and safely, further improving the user experience and building trust with organisations.

Advantages of Speech Recognition

Convenience

Convenience allows consumers to be more productive with their time without spending additional time on smaller tasks. Also, in some cases consumers do not need to carry or directly interact with their smart devices to initiate conversations with speech recognition technology, such as smart speakers, allowing them to work many tasks simultaneously.

Automation

Automation reduces time and money costs for businesses. By providing 24/7 customer service with no wait times and offering hands-free solutions to various tasks for increased productivity and efficiency, businesses can focus on more value-added activities. In addition, they are able to improve communication with their customers, as they can learn more about them through this technology and help provide tailored responses in a timely manner.

Personalization

Personalization creates more engaging and tailored conversations and experiences. Speech recognition technology can save each person’s voice biometrics, learn their habits, and provide specific responses to each query.

Security

Improved security helps industries such as banking, insurance, and health, as each person’s voice can serve as a unique identifier by taking advantage of characteristics like pitch and dialect, reducing the risk of fraud and theft.

Disadvantages of Speech Recognition

Errors

Errors and inaccuracies remain an issue that results from speech technology misinterpreting human voices. How fast people talk or pronounce words, their pitch and dialect, and the level of background noise generated from their environment can negatively impact the technology’s ability to understand what humans are saying.

Privacy

Voice-recorded data is always saved to better train speech recognition technology, often without the user’s consent, to better tailor each user’s experience. As a result, this opens up privacy concerns and whether users should be able to choose what to let the technology keep and discard or not. In addition, many of these technologies are always listening in to human conversations and picking up what we say to improve responses in the future, again often without the user’s knowledge.

Multi-Language Support

Support for all languages[24] is another major obstacle companies continue to work on. Currently, common languages such as English are supported whereas uncommon languages require more work before many countries can see mass adoption of speech recognition. It is difficult for companies to account for and try to interpret every language and dialect, as it requires very large sets of data to improve the technology and fine-tuning existing algorithms after additional research, which comes at a cost.

Required Training

Extensive training is required before speech recognition can function to its full potential. At the start, speech recognition works at a general level, where responses are the most accurate and personalised towards the users. However, once it has been trained over time through constant use, the technology begins to operate better. Therefore, it is on the user to always be using speech recognition for it to improve.

Chatbot

Chatbot or “chatterbot” is a software used to conduct an on-line chat conversation via text or text-to-speech, it is designed to convincingly stimulate the way a human would behave as a conversational partner[25]. Chatbot is related to the concept of turing test. Turing test is a test where the user is being test whether they are interacting with human or a chatbot. A chatbot pass the test when the user cannot differentiate whether they are chatting with a human or a chatbot.

History of Chatbot[26]

1966: Eliza -> Eliza is a early natural language processing computer program created by Joseph Weizenbaum[27]. The program works like an if-then statement, as it looks for keyword when providing the responses. The downside of this chatbot is that it is not conversational, it is more like a rule-based chatbot. The system answers to the questions provided by the users by matching the keyword to the answer bank on the system.

2000: Alice. Alice stands for Artificial Language Internet Computer Entity. This type of chatbot is designed to use pattern recognition to interact with the users. This algorithm is the foundation for most chatbots that we have today.

2010: Apple siri and Home Assistants. Apple siri was the first chatbot to serve as a human assistant that gives users a speech customer experience. More home assistants such as smart speakers was being introduced in this year as well.

2016 until present: AI boom. This era where people integrate more sophisticated chatbots and AI into their daily lives. For instance, the rise of the use of virtual agents.

How does a Chatbot work?[28][29]

Chatbots uses 2 core technologies to function which are machine learning and natural language processing. The use of machine learning in chatbots means that the machine continuously learns about the user’s behaviour and keep learning to be better by using the information entered by the user. The natural language processing (NLP) is a condition needed where the machine has to convert the human language or other information that the user enter to match the program’s language. Chatbot is connected to a cloud database and this NLP is needed to match the users’ language and the information on the cloud database. To illustrate this point, when there is an input from a user, the system analyze the request using the NLP, send the information to the cloud database, match the information is and what is available in the database, then send it back the response to the users.

Applications of chatbot[30]

Retail and e-commerce

Chatbot is useful for this industry because it can generate the availability of the product and the price as well as send the notification to the customers. It can also serve as purchase assistance by assisting the purchasing order and add personalized recommendation based on the purchase history. Furthermore, it can help with the post-purchase customer service experience by order tracking, refund or cancel requests.

Travel and hospitality

Chatbot is a huge help in the travel and hospitality industry when it comes to reservations and bookings. People can make the reservations online through website to reduce wait lines on the phone. It can also provide refund and reschedule assistance.

Banking, finance, and fintech

Customer issues in banking and finance often require immediate attention. Chatbots provide fast and accurate responses, making them increasingly popular in this space. In fact, experts predict 90% of customer interaction in banks will be automated by 2022[31]. Chatbot can help with personal financial information. Users can obtain their account balance and transaction details by interacting with chatbot. With the smart AI developed for chatbot, it can assist the loan and banking services. Customers are able purchase insurance online as well as receive wealth management services through chatbot.

Healthcare

In the healthcare industry, Chatbot can be helpful by providing conversational self-service. With the new online telemedicine websites and tools, patients can consult to a doctor using doctor-bots. It can also assist with appointment booking or appointment rescheduling and cancellation.

Media and entertainment

Chatbots have been deployed by news channels and publications to create better experiences for the reader. It can provide suggestions for on-demand content, it can give the recommendation based on the popular movies and the content that the users like to watch.

Education

Chatbot is proven to be useful for student assistance. It can act as a virtual advisor when asking about simple questions or about the library or any outstanding fees.

Advantages of Chatbots[32]

Increase customer satisfaction

Chatbots is proven to increase customer satisfaction because it can operate 24/7 as well as answer to multiple inquiries at the same time. This can help further reduce the wait times. As the chatbot is already programmed, the answer is consistent and can handle multiple languages. Moreover, with chatbot, it can add a touch of personalization by engaging customers with one-on-one conversations, maintaining a natural-sounding tone, and by being good at interactive communication.

Minimize cost[33]

It can help to reduce operational cost in the long run when installing chatbot. Operational expenses are ones that are required to run a business, such as labour and machinery costs. In customer service, operational costs include expenses related to hiring and training agents. By implementing chatbot to handle the customer inquiries, it is a one-time investment cost in the beginning, and overtime, it will lower the cost as it can cut down the expenses to hiring and training agents.

Analyze customer data

Chatbots are able to gather and store the customers' data to provide personalization when interacting with the users. Therefore, these data, for instance, sales data, can be used by the businesses to predict future sales or seek for sales pattern.

Disadvantages of Chatbots

Security concerts[34]

​​Chatbots deal with a lot of data collection. When you collect your audience data, it’s the company’s responsibility to keep it secure. The data needs to be transmitted from the chatbot to the cloud database in a secure manner. It must also be stored securely and only relevant data should be collected from the users.

Requires a lot of data

To understand the customer preferences, the chatbot needs data from the customer. It requires a huge amount of data to find the pattern or behaviour of the users. Therefore, data collection might be an issue as some people might not be comfortable sharing their information.

Influence the way human interacts[35]

As human beings, our brains have an inherent tendency to prefer simplification over complexity[36]. A Chatbot does not need emotional involvement during interaction. This simple yet repeated interaction can trigger cognitive laziness. Cognitive laziness is a limited mental focus when it comes to processing the information you receive[37]. Repeated interactions with chatbots trigger the constructions of a new mental model that will inform these interactions. It will be experienced as a different state of mind from which we interpret social interactions. Therefore, this can make our brain to not think critically which can affect the critical thinking process that can lead to being not able to differentiate true or false information.

The Future of Chatbots[38]

Chatbot is necessary for global growth. The global AI in telecommunications market is worth US$ 918.6 million in 2022 and is expected to worth US$ 10,399.9 million by 2032. Moreover, the spending on virtual assistance reached $3.5 billion in 2021 and 50% of knowledge workers will use a virtual assistant by 2025

Facial Recognition

Facial Recognition
Facial recognition is a biometric identification technique where the characteristics of an individual's face are analyzed to identify them [39]. A majority of facial recognition systems work by comparing facial features to a database of previously collected faces. The algorithm searches for a match for the system to identify the individual.[40]. However, if there is no match and the face print is not in the database, the individual cannot be identified [41].

Functions of Facial Recognition

First AI uses face alignment and detection where a machine learning algorithm locates the face in a video or image [42]. Most cameras nowadays have a built-in face detection function as popular social media applications such as Snapchat use it to add effects to photos and videos [43].. Once a face has been aligned and detected, the facial features need to be extracted and measured [44]. A Convolutional Neural Network (CNN) is a deep learning algorithm that can differ various facial features from one another and assign importance to them (Saha, 2018). The CNN extracts facial features from the image or video, which are used to identify an individual. A person’s face is broken up into multiple data points to measure one’s shape of nose, chin, eyes and other features [45]. An advanced algorithm is able to detect more distinct facial features such as the height of one’s cheekbones and the distance between the eyes. AI deep learning algorithms search on those data points and look for similarities to then match the extracted features with faces in the database.

Training AI for Facial Recognition

For AI facial recognition to become accurate in its findings, the system needs to practice on a large database of photosets [46]. AI models should be trained with facial images that vary in ethnicity, age, lighting, and angles as the face is not always perfectly positioned as a frontal headshot.

Applications of Facial Recognition

Face ID

Phone FaceID
A common use of facial recognition today in society is unlocking phones with Face ID.

Airport travel

Facial Recognition Kiosk
Facial recognition is used for efficient travel throughout the airport. Kiosks can be used to check-in an individual. The Vancouver airport is allowing Nexus cardholders the choice to use facial recognition through the Kiosks [47]. Nexus kiosks identify passengers utilizing facial recognition which is replacing the use of iris scanners in the kiosks [48].

Airlines such as Delta have a program available for sign up to have TSA checks where one’s face can be scanned instead of showing physical ID and a boarding pass [49].

Casinos

Casinos are now adopting the use of facial recognition for security and customer service purposes. As individuals can be banned from casinos, to keep track of black listed individuals casinos are utilizing facial recognition to identify and escort them out. Only relying on human intervention is difficult in heavily crowded spaces [50].

Privacy Concerns

To gather mass amounts of facial images to train AI facial recognition in its accuracy, companies sometimes gather photos on the web without people’s consent.

Facebook lawsuit

Major companies such as Facebook have been under several lawsuits for gathering people’s facial images for data. In 2017 Facebook lost the lawsuit and had to pay a settlement to the people of Illinois [51].

Face++

Face++ is facial recognition technology made by Megvii which is a company based in Beijing that is one of the leaders in facial recognition technology. Face++ age and gender recognition has achieved a 99% accuracy with a deviation of 5 years when present in real-life situations [52]. As for its face attribute recognition being at 97% accuracy in recognizing individuals despite changes in hairstyle, eye colour, and glasses [53].

Reliability of Facial Recognition

Inequity of Facial Recognition

Not all facial recognition systems are accurate and reliable in its findings. Even well-trained AI facial recognition systems can provide false positives as they can lack real-world context training. With people wearing face masks, sunglasses, and other accessories we as humans may still recognize faces, however an AI system might not as it depends on the level of training it has received. Facial recognition algorithms state to have a high classification accuracy, but a majority are in ideal scenarios.

There is an inequity in facial recognition algorithms with its accuracy as it does not apply to all demographics of people [54]. Several researchers have been conducted to expose the occurrence of errors between the different ethnicity groups [55]. Consistently the poorest accuracy in facial recognition are females with a darker skin tone between the ages of 18-30 years old. As facial recognition has uses for security purposes, this is a detrimental issue and causes concern in ethics.

Wrongful Arrest

There have been cases of individuals being wrongfully convicted and arrested of crimes they did not commit. In 2019 Nijeer Parks was arrested due to the use of facial recognition even though he was innocent. [56]. Police found a driver's license at the crime scene and facial recognition technology provided a match to Nijeer Park's face [57]. Without any other evidence the judge of the case signed a warrant for Nijeer Park's arrest because of the results of the facial recognition technology. As facial recognition has flaws in its accuracy, it should not be the only type of evidence to warrant an arrest on an individual. Police need to gather more evidence of the crime as well as take into account one's alibi before an arrest.

San Francisco bans facial recognition

San Francisco bans the use of facial recognition software by police and other agencies. Being the first major American city to ban facial recognition, despite being a tech driven city. Currently the international airport and ports that are under federal jurisdiction are allowed the use of facial recognition as they are not impacted by the legislation [58]. There is the fear of the technology’s potential abuse by the government. However, there has been backlash to this decision as some believe that the city should regulate facial recognition instead of an outright ban [59].


How likely is people to include AI in their daily lives

Benefits

Efficiency

The application of Artificial Intelligence in daily lives can better divide the work labor with humans, with AI not only being the assistant of us, but as substitutes to some designated job duties. This results in better allocation of labor as workers can be assigned to another job, enabling higher efficiency in work output. The cost of training can be greatly reduced as one AI machine is able to work 24/7, non-stop without ‘taking a break’, ‘sick leave’ or ‘annual leave’.

Accuracy

With the input of data, AI can execute tasks with more than 90% of accuracy. Such a high level of accuracy enables it to become part of our daily lives, that people need not to adjust themselves and cover for the mistakes that AI could have made. People can rely and trust on the application of AI given its high level of accuracy.

Provision of another living style

Humans have lived a long life without the support of advanced technology. The application of AI enables another new style of living given the aforementioned examples of how humans adopt AI into daily lives. Such a new way of living can provide an environment with a more variety of culture.


Concerns

Privacy

As AI relies heavily on the data input into its program, it may face the threat from cyberattacks for extracting the data. Since there is not a well-developed coping mechanism for AI to prevent such cyber security menace, concerns over the privacy of data may arise.

Job Elimination

It has been discussed that AI will play a huge role in the workplace for the coming years. With that being said, many occupations may be replaced and result in unemployment, especially jobs requiring relatively lower levels of skill. It has brought to the general public’s concerns that AI will replace approximately 85 millions of jobs by 2025.[60]


Authors

Hussein Fawaz Velika Primetta Jasper Chen David Tsoi Lok Long Siu
Beedie School of Business
Simon Fraser University
Burnaby, BC, Canada
Beedie School of Business
Simon Fraser University
Burnaby, BC, Canada
Beedie School of Business
Simon Fraser University
Burnaby, BC, Canada
Beedie School of Business
Simon Fraser University
Burnaby, BC, Canada
Beedie School of Business
Simon Fraser University
Burnaby, BC, Canada

References

  1. https://www.ibm.com/cloud/learn/machine-learning/
  2. your https://www.netsuite.com/portal/resource/articles/inventory-management/warehouse-automation.shtml#:~:text=How%20Does%20Warehouse%20Automation%20Work,tools%20like%20inventory%20management%20software.
  3. your https://www.netsuite.com/portal/resource/articles/inventory-management/warehouse-automation.shtml#:~:text=How%20Does%20Warehouse%20Automation%20Work,tools%20like%20inventory%20management%20software.
  4. https://aws.amazon.com/solutions/case-studies/amazon-robotics-case-study/
  5. your https://www.wonolo.com/blog/pros-cons-of-automated-warehouse-picking/
  6. your https://www.wonolo.com/blog/pros-cons-of-automated-warehouse-picking/
  7. your https://www.wonolo.com/blog/pros-cons-of-automated-warehouse-picking/
  8. your https://www.wonolo.com/blog/pros-cons-of-automated-warehouse-picking/
  9. your https://www.wonolo.com/blog/pros-cons-of-automated-warehouse-picking/
  10. your https://www.wonolo.com/blog/pros-cons-of-automated-warehouse-picking/
  11. https://yellow.ai/voice-bot/benefits-of-using-voice-ai/
  12. https://www.ibm.com/uk-en/cloud/learn/speech-recognition
  13. https://www.forbes.com/sites/cindygordon/2021/12/23/a-market-to-harness-speech-recognition-artificial-intelligence-ai-innovations-on-the-rise/?sh=5ad00502134d
  14. https://finance.yahoo.com/news/ai-speech-recognition-market-size-044900558.html?guccounter=1&guce_referrer=aHR0cHM6Ly93d3cuZ29vZ2xlLmNvbS8&guce_referrer_sig=AQAAALkeGRN5iQSFYt9t0uC_QL9V_67uQrBkmYgcisrBGkdTTVcFxETlQygTjzJjznE8Ra4ZpTwbU-hdbP-z_Z3WPIERjFuRc6LEG63FMLkwh1QWMhZDzy1uMLiggMRQx8rcp_qaPWMMsn0uF5uGzC2KAIACxG8svHW2mgoKmhIZTHj8
  15. https://www.forbes.com/sites/cindygordon/2021/12/23/a-market-to-harness-speech-recognition-artificial-intelligence-ai-innovations-on-the-rise/?sh=5ad00502134d
  16. https://www.zdnet.com/finance/paypal-enables-voice-activated-payments-via-siri/
  17. https://stories.starbucks.com/press/2017/starbucks-debuts-voice-ordering/
  18. https://www.cnbc.com/2021/10/27/mcdonalds-enters-strategic-partnership-with-ibm-to-automate-drive-thru-lanes.html
  19. https://www.businessinsider.com/mcdonalds-drive-thrus-voice-recognition-automated-chicago-2021-6
  20. https://www.hyundainews.com/en-us/releases/3263
  21. https://voicebiometrics.ai/blogs/voice-biometrics-use-cases/
  22. https://voicebiometrics.ai/blogs/voice-biometrics-use-cases/
  23. https://voicebiometrics.ai/blogs/voice-biometrics-use-cases/
  24. https://news.mit.edu/2021/speech-recognition-uncommon-languages-1104
  25. https://en.wikipedia.org/wiki/Chatbot
  26. https://www.youtube.com/watch?v=YnRrdHsV1xE
  27. https://en.wikipedia.org/wiki/ELIZA
  28. https://www.youtube.com/watch?v=o9-ObGgfpEk&t=306s
  29. https://www.helpscout.com/blog/ai-customer-service/
  30. https://freshdesk.com/customer-service/chatbot-applications-blog/
  31. https://www.smallbizgenius.net/by-the-numbers/chatbot-statistics/#gref
  32. https://www.google.com/url?q=https://www.revechat.com/blog/chatbot-business-benefits/&sa=D&source=docs&ust=1670313416234250&usg=AOvVaw2ANWqS8pf5IHIA-ZqK75iS
  33. https://verloop.io/blog/cut-costs-with-voice-chatbot/
  34. https://www.moengage.com/blog/pros-and-cons-of-ai-chatbots/
  35. https://www.cmswire.com/customer-experience/chatbots-the-illusion-of-companionship-without-the-demands-of-friendship/
  36. https://www.cmswire.com/customer-experience/chatbots-the-illusion-of-companionship-without-the-demands-of-friendship/
  37. https://exploringyourmind.com/the-connection-between-cognitive-laziness-and-social-media/
  38. https://www.google.com/url?q=https://www.analyticsinsight.net/top-10-chatbot-trends-and-predictions-to-know-in-2023/&sa=D&source=docs&ust=1670313637625429&usg=AOvVaw1R4PUqiSCZXbaH86ihNZSH
  39. https://www.pxl-vision.com/en/blog/machine-learning-and-how-it-applies-to-facial-recognition-technology
  40. https://www.pxl-vision.com/en/blog/machine-learning-and-how-it-applies-to-facial-recognition-technology
  41. https://www.pxl-vision.com/en/blog/machine-learning-and-how-it-applies-to-facial-recognition-technology
  42. https://www.pxl-vision.com/en/blog/machine-learning-and-how-it-applies-to-facial-recognition-technology
  43. https://www.pxl-vision.com/en/blog/machine-learning-and-how-it-applies-to-facial-recognition-technology
  44. https://www.pxl-vision.com/en/blog/machine-learning-and-how-it-applies-to-facial-recognition-technology
  45. https://recfaces.com/articles/ai-facial-recognition
  46. https://recfaces.com/articles/ai-facial-recognition
  47. https://vancouversun.com/news/local-news/yvr-to-become-first-canadian-airport-to-use-facial-recognition-at-nexus-kiosks#:~:text=from%20our%20team.-,YVR%20to%20become%20first%20Canadian%20airport%20to%20use%20facial%20recognition,iris%20scanners%20in%20Nexus%20kiosks
  48. https://vancouversun.com/news/local-news/yvr-to-become-first-canadian-airport-to-use-facial-recognition-at-nexus-kiosks#:~:text=from%20our%20team.-,YVR%20to%20become%20first%20Canadian%20airport%20to%20use%20facial%20recognition,iris%20scanners%20in%20Nexus%20kiosks
  49. https://www.youtube.com/watch?v=vUd9S-6uhB0
  50. https://www.youtube.com/watch?v=vZIJbjWXskY
  51. https://www.youtube.com/watch?v=9VW8J08LcZo
  52. https://en.megvii.com/technologies/face_recognition
  53. https://en.megvii.com/technologies/face_recognition
  54. https://sitn.hms.harvard.edu/flash/2020/racial-discrimination-in-face-recognition-technology/
  55. https://sitn.hms.harvard.edu/flash/2020/racial-discrimination-in-face-recognition-technology/
  56. https://www.youtube.com/watch?v=nGStQVeCYuw
  57. https://www.youtube.com/watch?v=nGStQVeCYuw
  58. https://www.nytimes.com/2019/05/14/us/facial-recognition-ban-san-francisco.html
  59. https://www.nytimes.com/2019/05/14/us/facial-recognition-ban-san-francisco.html
  60. https://www.weforum.org/reports/the-future-of-jobs-report-2020/
Personal tools