Technological innovations give rise to both aspirations and anxieties. Popular acceptance comes when the benefits from an innovation outweigh the damage. This happened in the case of the computer. However, AI is a different devil; it can cause serious, irreparable damage. That is why AI governance, with the help of a strong framework of laws, treaties, rules and appropriate policies, is very important. The following areas need particular attention.
Data collection
A single sample may be sufficient for a human being to understand how a particular work is done. But AI needs large amounts of data to work flawlessly. Using the data, AI and its models train themselves to respond in a given scenario. So, a country that allows unrestricted collection of data from its subjects will have an edge over countries that have legal restrictions on data collection.
Thus, a set of guidelines or treaties with all nations as parties to them is necessary to ensure a level playing field and to reduce the chance of companies resorting to unethical practices. These should cover insurance, stock markets, copyright, trademark, patent, etc., where AI-powered software can wreak havoc. Software can be designed or altered using AI in such a way that it favours certain industries, companies, sectors, etc. All this can be done without making the user suspicious. The software developer may benefit, but the user and the nation will suffer. In a fast-growing economy like India, the Securities and Exchange Board of India (SEBI) and other regulatory agencies have to be equipped to handle these problems.
At present, governments across the globe seem to understand data protection only as protecting the data of natural persons. The European Union’s General Data Protection Regulation (GDPR) forgets that the data available with a legal person like a company or a large conglomerate is equally, or sometimes more, valuable.
A country’s growth depends to a large extent on its industry and business. Companies engaged in these activities record the problems they have faced and the solutions they have found. This data is of immense value, particularly in the era of AI. This is the reason why companies of some countries go on a takeover spree across the world. The data — research materials, patents, problems faced, troubleshooting methods, etc. — can be processed later using AI to make and sell products which can be claimed as being different in legal terms.
Once AI reaches higher levels, it will be able to replace humans in certain areas. Companies belonging to unfriendly countries can create heavy job loss in a country, which will lead to a chain of unfortunate events unless its government decides to take drastic measures in order to rectify the situation. Such methods can also be used to overthrow governments. The blame would be put on AI but the reality may be far from this.
Application of law
Several countries are still in the process of framing AI policies and laws. So, companies have to depend on contracts. No need to say that the contracts should be drafted with extreme foresight to cover all legal problems that might arise in future. Legal disputes not only consume valuable time and money but also slow down the whole system that develops AI. The courts may interpret the cases as per their understanding of AI, a complex and relatively new area for them.
Fixing liability
These days one comes across videos, photos, documents, etc. which often show celebrities and other influential personalities making surprising, sometimes shocking, statements. On fact checking one finds out that they are AI generated. If such issues are not addressed promptly, they can lead to serious problems, between not just persons but also nations. In grave crimes like these, liability has to be fixed.
It is not possible for a single individual to create complex software capable of making such products but companies can because they have the necessary resources at their disposal, like experienced employees, experience in making software, etc. Putting restrictions on these companies may not be easy but not impossible.
It is a business practice among companies to outsource work. At times companies that outsource work may not reveal the full nature or purpose of the job. In this case who will be held liable if the product is used for unlawful activities?
There must be laws to govern such unlawful activities and protect people who may not know the real purpose of the job handed over to them. Thus, the company that ordered the job concerned and the involved parties who knew its actual purpose or is capable of understanding the purpose in the course of work should be held liable, not companies or persons who are not capable of ascertaining the purpose of the work.
Jurisdiction
Another issue that needs to be considered in order to apply the laws is jurisdiction. A country can make laws to rule its own territory, but not its neighborhood. This will certainly become a difficult problem when AI becomes widely used. AI software can be used to collect data from a website whose content has copyright. If it is done from a country that does not have any AI law, then there will be a tricky situation.
Copyright violation
Copyright violation is another certainty as AI cannot proceed without data. Data may be taken from sources that are copyright protected, and AI may not mention the source. Even if AI does so, it would use its own intelligence to present the matter in a way it feels is right. There may be thousands of such cases in the market. In such a scenario, against whom will you file a case?
Against the user or the developer or the company that provided the AI service? Because the user may not know how the AI software works or how the query is processed by AI, the developer or the company may be held responsible. But they may claim that AI analyses the user’s inputs, past searches, etc. and alters its working style on its own. So, this would lead to a never-ending blame game. The courts too will find it difficult as so many similar cases would be piling up and clubbing them would be difficult as there may be differences in the key issues.
There are instances where lawyers have used AI and AI hallucinated and created non-existing cases. If the court uses AI to dispose cases, the result can be the same. But the court would claim legal immunity in case of a mistake. Thus, in order to prevent these there must be clear laws even preventing the court from shifting the blame to AI. The court too has to do due diligence like the lawyer.
So, there must be an internationally accepted framework, preferably a treaty, where guidelines are put in place for AI development. As AI development has the potential to interact with data outside a specific jurisdiction, it would be a safer bet if every country abides by a specific standard. Importantly, the process through which courts admit AI cases and AI-related issues, to what extent courts can use AI, and the procedure courts should follow may be formulated.
Threat to national security
Some Indian YouTubers were arrested recently for spying. However, with the help of AI you
can be used as a spy without your knowledge.
We look at autonomous cars, ADAS systems, etc. with admiration and approval. But how many people know that these systems often use Lidar or other advanced sensors and software that are capable of collecting enormous amounts of data? Lidar sensors can create a 3D image of what is ahead of the car and work as programmed. That is why the car stops if it “sees” an object in front of it. The sensors are also capable of identifying the road condition, potholes, undulations, buildings, etc. as they scan the road surface. The data collected can also be used for purposes that are harmful to the country, especially in the event of a war.
Several countries are developing their national highways in such a way that they can be used as emergency runways for aircraft, including fighter planes. If data from Lidar and other sensors are transferred to an unfriendly country, it can determine, by using AI, to study the roads’ maintenance history, pothole condition and proximity to aviation fuel stations and airbases, etc., which place, national highway or road would be chosen to land fighter planes.
As most of these sensors are made in China, it would be essential to transfer the data collected to the Chinese company in order to improve or develop the sensor. The majority of automobile manufacturers have collaborated with Chinese companies to make such sensors for them. Toyota has collaborated with RoboSense, a Chinese company, to make the Lidar sensor for their cars.
Many modern cars come with the over the air (OTA) update feature where their software can be updated without visiting the dealership. The data once collected can be uploaded to the company’s server by even misleading the user. The system may show downloading but actually what might be taking place is uploading.
One can ask whether satellite images are not good enough. Maybe. But an image taken from ground level is different. That is why we often look at objects from different angles to identify imperfections. So, it is essential that there must be laws in place to prevent certain data they store from exiting the country.
Vehicle dashcam with Wi-Fi and app functionality
Even the dashboard camera, or dashcam, used in cars can pose serious security problems, especially the aftermarket ones. Several dashcams available in the market use apps and Wi-Fi to transfer or access recorded videos. This means the creator of this dashcam or the app has direct access to the content.
Some dashcams come with the remote access feature, engine cut-off function, etc. which enable the user to access the video even from a faraway location or interrupt the functioning of the engine. They may be convenient features. But in view of the security threat they pose in the AI era, the government needs to keep a close watch on them.
AI and Job loss
Among the first things that come to one’s mind at the mention of AI is job loss. To an extent it is a reality. But to what extent is the question.
When computers hit the market or the Internet was made available to the public, the widespread feeling was that many jobs would be lost. In the event these opened new avenues. Only time will tell what AI will do, for AI is designed to think on its own, make decisions, implement them, and review its actions, that too at unprecedented scale and speed.
But one area AI will open up as soon as it is adopted widely is jobs to oversee its actions. It has often shown a tendency to go overboard, and engaging another AI software to monitor it may create more complications. So, it would be wise to have human intervention as and when needed.
One fact offers some comfort: AI lacks consumer power. What it needs is data, electricity, computers, hard disk space, etc. What is feared is massive job loss and heavy fall in consumer income. Then, if you do not have a buyer at the other end, why deploy AI to make a product?
All said, the future AI promises is as dangerous as it is exciting. Here is a possible AI product of tomorrow: the leader of a powerful country being shown as threatening to use nuclear weapons against a rival power. Imagine what will happen next.
And can anybody rule out the possibility of an AI product beginning to think independently and deciding to defeat the human race, which it thinks is inferior?
Going by the current AI excitement, every sensible person must be wondering whether fear of the unknown, the most powerful survival tool of all animals, has taken a back seat in the human DNA.
Comments