AI has been consistently at the forefront of technological innovation. With all the recent incidents regarding personal data collection and manipulation, we have realized that the majority of infrastructures have yet to evolve in order to accomodate a society embedded in the Internet of Things (IoT). To ensure a smooth integration on AI, we have decided to map out the new generation of IoT in order to research and address the fundamental errors that AI will currently face.

We are a group of students from the University of Waterloo with an intense desire to learn about the resistance and adoption of AI in our current society. We have collected data on the topic through primary research (limited by COVID-19), news & industry articles and academic papers. Together, we explored the different ways that AI is being both accepted and rejected, and any existing solutions to address the latter concern.

What Is Artifical Intelligence?

The world is becoming more technologically advanced at an exponential rate. Smartphones, Bluetooth, and the omnipotent Internet of Things have connected us in ways never before imaginable, and as a result, the world has globalized. The next big leap in human-created technology is to utilize the robotic systems that we have crafted in order to serve us as efficiently and effectively as possible. Artificial Intelligence (AI) is “the ability of a computer program or a machine to think and learn” [1]. AI enables machines to learn from experience, relying on deep-learning and natural language processing [2].

Why Research AI?

AI is central in the discussion in emerging technology and it is one of its most intricate applications. However, the question remains, what makes AI so important? As of now, all new creations were built upon existing infrastructure according to existing rules set by governments. However, none of this has been rebuilt which raises the question, how can a society that has yet to be changed to accept technological advancements adapt to the inevitabilities of evolution?

Technology separates the logical and emotional/moral point of view of thinking [3]. AFor example, logical thinking represents AI and emotional thinking represents humanitarian considerations. Most rejections of AI are distilled by fear, specifically caused by the lack of knowledge and experience with AI. Arising issues like data privacy are forged by the unregulated powers of tech companies which is a consequence of the government failing to acknowledge and pushing back on the evolution of technology. Everyone knows AIs exist but it is not a part of everyday life, and for acceptance to be ensued, measures must be taken to address these concerns.

Research Methods?

Our research was compiled through:

• Surveying over 20 people across Ontario
• An in depth interview with an industry expert
• Academic journals, government documents, data sets, news articles and industry articles

Infrastructure

As AI integrates into society, proper infrastructure will be required to support it. At the most basic level, infrastructure (i.e. think: roads, buildings, cities, transit systems) in its physical forms needs to be well-developed. For example, to effectively support autonomous vehicles (AVs), we need properly designed road systems to make them viable and safe. This could mean having sensors on roads or street signs to send signals to other vehicles to help cars navigate city streets. [4] Additionally, having exceptional networking systems, such as 5G, will be crucial for AVs in the future as it will allow for faster data transfer rates between cars and the environment [5]. Faster data transfer will enable AVs to send and receive key data to make safe decisions.

Another area where infrastructure can be optimized is in hospital settings. The medical field has made massive advancements with AI, and technologies such as medical imaging diagnostics and virtual health assistants, [6] but only when hospital rooms are equipped with the space and equipment necessary to support these can it be implemented to its fullest potential. These applications exist among many others. Without proper infrastructure, there is no artificial intelligence.

Environment

AI is most advantageous to us when applied to large, dimensional and unstructured datasets [7] – and this is where most data on climate, weather and other long and short-term environmental factors are found. AI is a relatively new tool in the environmental sector and it’s being used to manage the impacts of our own activity on the planet. There are many specific areas of application, but the following have been identified as particularly pertinent:

Autonomous Vehicles (AVs) are not only self-driving, but are also interconnected, capable of sensing their environments and making decisions accordingly [8]. The primary benefit to an AV-lead traffic system is a significant decrease in urban CO [9] emissions – according to the Institute for Transportation & Development Policy, this could reach up to 80% worldwide [10]. A major roadblock in the implementation of AVs, and in developing self-driving cars, is the fear of loss of control and poor safety, despite the convenience and potential safety benefits offered by AVs [11].

Smart agriculture and food systems have massive potential for slowing the environmental repercussions linked directly to producing enough food to feed the earth’s population and its increasingly unsustainable preferences. AI could automate data collection, provide early detection of crop diseases and optimize agricultural inputs [12]. Ultimately, it would increase resource efficiency. Sources of resistance to smart agriculture are mainly legal and regulatory [13].

Other areas where AI can be implemented for the betterment of the natural environment is in wildlife tracking and therefore preservation, in weather and climate prediction leading to smart disaster response, and in the creation of intelligent, connected cities that use data to manage assets, resources and services efficiently.

Education

Recently the landscape for education has evolved drastically. Considering the development in the current compositions of families, schooling options, wealth disparities, and the ever-changing economic demand for new skilled workers and technology, it is surprising to see that there has been minimal change in the way we teach [14]. In a world where we are constantly improving, how is it that we haven’t improved the way we teach?

What must be realized is AI is not going to be the new teacher, rather an assistant. This can revolutionize education by opening up new opportunities to teach anyone, anywhere. AI assistance can fill the gaps in learning and teaching by providing customized feedback. AI can improve efficiency and streamline administrative tasks such as grading homework, assessing written responses, allowing more time for teacher-to-student interaction [15].

Additionally, AI can provide universal access to education for all students. An AI tool called Presentation Translator is a free plugin for PowerPoint that can create real-time subtitles of what the teacher is saying. Duolingo is an excellent example of AI cooperation [16]. This app allows 300 million users to enjoy 90 courses in 22 languages. To continually improve their product, they use AI to uncover new insights on language structures and how people learn.

A major roadblock in widespread adoption of AI is the fear that it will take over. By rejecting AI, we are prolonging inevitable technological advances, depriving future generations. We conducted a survey on people’s experience with education and AI, to gain insight on how the presence of technology in education affects current views on advancing technology. The responses were disappointing, with many individuals claiming that “[Educational curriculums] teach with content from the 90s”, “Much of the technology used in the education system was outdated and irrelevant.”, and that “I’ve succumbed to learning myself.” These sentiments do not reflect the level of technology we’ve developed for education - it is clearly being under-utilized.

Socio-Economics

Slowly, as the benefits of smart devices are being recognized, markets begin to adapt to new technology. With the digital infrastructure currently in place, AI aims to cultivate an efficient lifestyle. However, it is the tech companies and businesses that hold the most responsibility with AI; they need to produce products and services that humans can actually use.

Specifically, the automation of labor is one area that has a noticeable economic impact. AI has the potential to grow GDP by 26% by 2030 and on a larger scale, the adoption of AI is capable of bridging the gap among developed and developing countries. The main political argument that deters adoption of AI is the inevitability of job severance and the risk that developing countries will fall behind [17]. However, it is proven that an increased demand for handmade products will be a result of economic change. There will be more opportunities for work within fields such as data privacy and user experience, ensuring the economic cycle continues [18].

As of 2018, there are 4,635 AI companies globally [19] who are paving the way for new technologies and industries for which AI can be adopted. These companies aim to save consumers money, time, effort, and also to educate the public on the benefits of AI through their products. To understand the current industry, we highlighted a few companies that are focused on the goals mentioned above. It is also important to examine the ethical and privacy issues surrounding AI.

Saving Consumers Money

Hopper is a company that leverages AI through machine learning to save consumers money. They use large datasets combined with machine learning engines to predict the right time to purchase plane tickets [20]. Airlines have open APIs which companies like Hopper can use to track flight-price history. Using these data points they forecast what the future price of a current plane ticket will be, and recommend to consumers when they should purchase the tickets [21]. Additionally, if a consumer is flexible on when they travel, the machine learning model can recommend flying on a different day in a similar time window [22].

Saving Consumers Time

Companies like Uber and Lyft use AI to provide consumers ETAs on rides and food delivery [23]. These efforts allow the consumer to feel confident in the service they are receiving and allow transparency with the consumer. The main goal of Ubers efforts is to reduce the need for ‘surge pricing’, and to predict rider demand, by implementing machine learning [24].

Additionally, Google Maps and their acquired company Waze, save consumer time by providing AI-predicted traffic updates. These predictions allow for users to plan their trips days in advance. Google Maps uses anonymized location data from smartphones to track traffic speed [25]. In acquiring Waze, Google gained technology to incorporate user-generated data on construction and accidents [26].

Saving Consumers Effort

Wave is an accounting platform for small businesses. They use large numbers of transactions to complete mundane tasks, which users normally do themselves, by means of machine learning [27]. Categorization is one area where they can do this [28]. When new transactions are added to an account a user usually must categorize these transactions into different areas such as “Meals and Entertainment” or “Office Supplies”. By leveraging past transactions, Wave can predict how future transactions should be categorized [29]. This saves the user the hassle of logging in every time a transaction is added and manually deciding how it should be categorized [30].

Privacy and Ethics

The biggest safety risk associated with using data used to power AI and Machine Learning [31] is in privacy. Algorithms can now replace human insight by using millions of existing data points to predict future ones [32]. But where data points come from, and whether or not the algorithms are taught to sway their suggestions to benefit corporations, are points of contention. Companies like Google are able to use your search history to predict your preferences. But what if you wanted to persuade a user to buy the new Google Home? You could target prospective buyers and strategically place ads to entice them to buy from you. Now re-read the last sentence but replace ‘buyer’ with ‘voter’, and ‘buy from’ with ‘vote for’, and now you’ve essentially used machine learning to win an election [33]. The voters will never know that you’ve had an effect, but politicians can essentially buy votes.

This is what happened in the Cambridge Analytica user privacy data breach [34]. If you’re a smaller company, but still want to leverage a machine learning model, you will have to buy data from bigger companies, or data aggregators. This creates ethical issues of selling data. No one should be able to profit from your information, which is a potential problem with AI. The method to overcome these concerns are to enforce Privacy by Design [35], and new legislation. Privacy by Design encourages businesses to take proactive action in regards to privacy, building it into their core values. Additionally, encouraging lawmakers to improve current regulations (such as PIPEDA [36], FIPs [37], GDPR [38], California Consumer Privacy Act [39] etc.) will enforce businesses to have privacy in their minds. The US Congress has a bill known as the “Fundamentally Understanding The Usability and Realistic Evolution of Artificial Intelligence Act of 2017” which aims to understand and prepare for changing times and most importantly, protect the privacy rights of individuals [40]. Enforcing these two methods will mitigate the fear around privacy and ethics in regards to AI.

Gap 1: Lack of Public Knowledge

Much of the resistance to developing AI and its implications stems from a lack of exposure and proper relevant education.

Gap 2: Fear of Data Collection

Another major source of resistance stems from the public’s fear that their data is insecure in the hands of the corporations or otherentities that use it to develop AI technology.

Our team was excited to explore a topic that we’ve never before seen articulated in an academic setting. It was interesting to navigate this system and approach the project with different perspectives, experiences and interests. What we did not anticipate, however, was the sudden onset of COVID-19 that has quickly transformed the world and sent it into an international pandemic. This pandemic developed quickly throughout our process, and halfway through it forced us to leave university and practice social distancing. This influenced how we could work on our project and how we could collaborate effectively with the task of working remotely.

Luckily, we were able to overcome most challenges by using personal computers, virtual meetings, collaborative design softwares and writing platforms. Unfortunately, we were unable to conduct certain primary research (e.g. in-person surveys, additional interviews) due to the distance, but attempted to compensate with online surveys. We acknowledge there may be some gaps in our research due to these challenges and hope to continue researching them in the future.

WIth COVID-19 on our minds, it's important to note that although we chose not to make the medical field a focus, AI has a significant role. The current situation did force us to think about AI as a potential tool in helping to slow the spread, treat and possibly cure COVID-19. Perhaps the largest challenge in fighting the virus, in Ontario especially, is the availability of testing. Many are claiming that the current “official” number of cases are far lower than reality, and that there’s a “backlog of more than 8,400 tests, with people waiting at least 4 days between test and result” [43]. AI, using deep learning algorithms, has made huge advancements in the field of diagnostics, making them cheaper, faster and ultimately more accessible [44]. Already, AI is being used to track the virus using machine learning, use computer vision to detect infection, employ robots on the frontlines, and to speed up drug research. DeepMind, Google’s AI research lab, “recently declared that it has used deep learning to find new information about the structure and proteins associated with AI” [45]. This is a process that, without AI, would have taken many months to complete.

This is, of course, just one of many potentially life-changing or life-saving applications of AI. But the current state of the world has certainly helped us to contextualize AI’s applications. It is our hope that, starting with our own generation, humans are able to overcome what seems to be an intrinsic resistance to artificial intelligence. We must recognize that this technology must be utilized in order for it to reach its full potential in benefiting humans, and whatever perceived loss of humanity we will incur through AI use will, in time, only reignite the arts that we have lost through a world obsessed with efficiency and progress.