What To Do With AI?

The federal government, states and cities think they know....Do they?

Instead of the typical monologue like list of federal, state by state and local (yes some cities are working to regulate AI) analysis of pending, passed or failed legislation I thought groups might be more insightful

While not every state is in the game yet (and only a few cities), one assumes all will be interested in regulating AI to some extent in the future. Why? Money, fairness and civil rights (for surprisingly many forward looking governmental entities) just to name a few motivating principles.

The minute this post is sent off to all of you, is the minute it is likely dated. Therefore, realize that passed, or pending or even failed legislation (yes AI legislation can rise again phoenix like) might be different now or in the future. The snapshot here is more of a map of the philosophical topography providing insight into what states are focused on and what they might be focused on in the future in the regulation of AI.

Is That Really Your Face?

I remember this insult from my brothers in the midst of backyard roughhousing. But, in AI world, a larger than expected number of states are already wary of government collection, storage and usage of facial recognition data. These states either having pending or passed legislation restriction the use of facial recognition. Alabama, Colorado, Illinois, Vermont, California, Georgia, Maryland, Massachusetts, New Jersey, New York, Rhode Island, Washington, Missouri, Nevada). Most of those states also considered legislation to create a government agency (and in some cases fund a director) regulating AI going forward. If you are doing anything with AI now or in the future, you should expect that your startup or company or department will have to refer to federal and state regulations regarding the adoption of AI tools. Just like departments of commerce, insurance, education, etc at the state level, the “department of AI” (might not be that un-creative of a name) will be something to contend with.

None of the legislation outlines the qualifications for a person to head these newly proposed AI commissions or agencies. The backgrounds of persons chosen for such roles offers some insight into the philosophy AI regulation will endorse. Some of the legislation establishes a commission (most are either expressly permanent commissions and others are unspecified, but knowing a bit about government they will likely be permanent) to advise government about AI policy. Likewise here no provided qualifications for commission members, but interesting collections of expertise can be imagined. A commission to advise the governor of X state on AI might have tech experts, professors, philosophers, folks with law enforcement experience, civil rights advocates, etc. AI is going to impact every facet of life now and in the future (see below) so a wide array of expertise and perspectives is likely the best construction for such commissions. We shall see what the various states decide on when those commissions are finally formed. Cities like Baltimore and some in Washington state are also outlawing or restricting the use of facial recognition software without the creation of commissions.

Some states are focused on nuances of AI and society, business, culture, etc. that are interests of those states’ residents. For example, California is already focused on analyzing deepfakes (likely influenced by the entertainment business interests in the state). A trend that might sweep the country, California also has some potent disclosure requirements regarding AI models.  The trend toward transparency of training data used for creating models, filtering and censoring of that data as well as the filtering and censoring of the output data will be battles to be fought by lobbyists and others for the rest of time no doubt.

New York, New Jersey’s and Nevada’s commission will focus on AI and its connection to the growth of the state’s economy.  Other states like Rhode Island and Missouri established commissions focused on the impact of AI on state government, hiring, promotion, etc.  Washington’s commission is to be focused on disparate impact of AI systems on its residents.

Mississippi has set its legislative sights on the mandating the teaching of technology and AI curricula in public schools. For the future of education generally, this would be a welcome trend in school curricula nationwide. (Says the tech nerd/lawyer).

Hawaii, unlike any of the states surveyed so far, focused its AI legislative initiative on encouraging companies to develop AI technologies by offering tax credits.

Michigan went more precise in mandating the use of independent AI experts to evaluate the fairness of algorithms used by state agencies such as the unemployment agency. Companies in the AI space should likely get used to requests for disclosure of AI algorithm training data, filtering and output data filtering. Citizens across the ideological spectrum are going to want to know what the AI algorithm is censoring and/or promoting.

Cities Weigh In

The cities of New York and Washington, D.C. have focused on auditing and regulating AI tools to avoid discrimination. The city of New York focused on hiring decisions and mandating candidate notification if AI tools are used in the hiring process. D.C. more generally is looking out for the “discriminator use” of AI algorithms. To tease out intentional design of an AI algorithm with a discriminatory purpose, companies making such algorithms will have to keep detailed records of the training data used and all steps involved in their modifying of the available data and output results. Companies engaged in such technological choices will have to increasingly make those choices with lawyers in the room glancing between relevant regulation and the technological reality of how the sausage of the AI algorithm was made.

The Octopus of AI

There is not going to be “the list” of places, events, organizations, etc that will be affected by the implementation of AI. Just a summary of some of the arenas provides plenty to ponder about the reach of the octopus.

Financial Tools, Digital Finance, and Transaction Solutions: Customers have the ability to register for credit cards, seek out loans, establish trading accounts, and obtain finance or investment guidance – all achieved without the need for face-to-face communication with a person. Artificial intelligence can take the lead in credit determinations, while automated advisors will eventually provide financial guidance. AI-enabled safeguards and fraud detection measures now secure transaction processes.

Insurance Coverage: The insurance sector relies heavily on underwriting and data modeling, creating an ideal environment for numerous applications of AI. Mirroring the financial services sector, insurance firms can employ AI for decision-making in underwriting. Additionally, AI allows insurance providers to gather data over periods, enabling a more fluid approach to insurance - such as smartphone applications that allow users to provide live driving information for AI-driven technology to determine premiums or rewards for safe driving.

Automotive: As AI continues to advance, it is expected to play an even greater role in the automotive industry, enabling vehicles to become even smarter and more capable. “Driver Assist” technology can make AI-powered decisions that alert inattentive drivers or take emergency action (like braking at intersections). Autonomous and self-driving vehicles will be powered by AI that operates vehicles in place of human drivers. AI is also increasingly used to improve vehicle safety, performance, and efficiency. For criminal law, the ability to access the data gathered by these censors will increasingly be used in investigation and criminal trials. 

Supply Chain Management: Much like the auto sector, the logistics field might witness sustained progression towards self-navigated delivery vehicles enabled by autonomous AI. These could be terrestrial delivery vehicles, unmanned aerial vehicles, or other modes of self-sufficient transport technology. Furthermore, AI is deployed to enhance the efficacy of the supply chain and refine vital logistics components like itinerary preparation and freight loading. Amazon built its entire empire on logistics and the transportation industry thrives shaving minutes and hours from transport times. The predictability of AI machine learning models will be the de facto decision makers now and in the future for this industry.

Healthcare & Biomedical Equipment: Artificial intelligence holds the capacity to dramatically transform the manner in which healthcare services are delivered to patients. AI can enhance patient care quality, lessen the workload on healthcare providers, and assist in preventing medical errors. AI can refine standard clinical procedures, such as facilitating accurate diagnoses, streamlining analysis of lab results or medical imaging, and managing patient care coordination. In terms of medical devices, AI can oversee health data to instigate automated interventions. Similarly, AI integrated with wearable technology can monitor health indicators like cardiac rhythm, physical activity, and sleep patterns to customize care and evaluate its long-term effectiveness. Those of us with various wearable devices already have access to many such health data points being recorded nearly continuously throughout the day.

Retail, Digital Commerce, and Accommodation Services: Artificial Intelligence equips retailers and service providers in the hospitality sector to foster a significantly individualized bond with customers. AI can curate personalized shopping experiences, both online and in physical stores, while also tailoring special deals, promotions, and potentially even pricing. AI can assist in fine-tuning product suggestions – envision, for instance, presenting your face to a store's kiosk and receiving tailored advice on makeup and beauty products.

Promotion & Publicity: AI is already a significant contributor to the digital advertising landscape, playing a pivotal role in campaign design, auctioning for paid media, and evaluating campaign success. As the advertising industry pivots away from cookies and individual-level surveillance, AI-enabled stochastic modeling might become more crucial for firms to strategize campaigns and evaluate their impact. AI also presents opportunities to dynamically tailor advertising messages instantly, potentially enhancing the relevance and efficacy of promotional campaigns.

Production: Many companies in the manufacturing sector has predominantly shifted to automated, robot-led assembly procedures. Could that continued development bring manufacturing back to the United States? AI holds the promise to automate even more segments of the production processes that still depended on human intervention despite the advent of robotic technology, such as quality assurance or safety examinations.

Media & Amusement: A good number of individuals might have noticed when their content providers' algorithms (such as those used by Netflix or Spotify) started accurately predicting their preferences. AI facilitates the delivery of tailored recommendations for visual or auditory content. However, AI is also empowering the actual process of content production. For instance, digital gaming has already introduced 'worlds' in which some character actions are partially generated in real time by AI. Text to video tools like runway.ml is in its infancy but attracting investment nonetheless. These revolutionary developments could expand to other forms of content and immersive experiences. The current writers and actors guild disputes with production companies has as its focus concerns about the encroachment of AI into content writing and acting.

Learning Sector: AI finds use in a multitude of contexts within the education sphere, such as in custom-tailored learning and systems aiding educators in lesson development and delivery. The necessity for distance-learning solutions amidst the COVID-19 crisis fueled exponential growth in the EdTech industry. EdTech persists as a domain where innovative AI solutions are being created for deployment in progressively digital classroom environments.

The Debate That Never Ends (Like Star Wars Sequels)

As with any regulation, companies trying to make a buck and innovate want certainty. For them, a web of competing, disparate and sometimes contradictory regulations between cities, states and the federal government is to be avoided. For the rest of us, governmental regulatory innovation flourishes when everyone who wants to weigh in (cities, states, the federal government) gets that opportunity. May the best ideas win so to speak. The tension here is not new in the field of AI. But, it will be interesting to see how the evolution of regulation either does or does not become consolidated at the federal level essentially closing all the regulatory experimentation that is already underway throughout the country. Where you end up here might well be based on where you sit (entrepreneur, civil rights advocate, investor, privacy enthusiast, etc). No matter where you are philosophically, the debate is tacitly underway as evidenced by cities adopting AI legislation instead of awaiting federal legislation or regulation.

Just as with any newly adopted technology, all the contours of what is acceptable, legal, what imposes liability, etc will be worked out over the next decade or so. These realities will keep AI educated lawyers quite busy and the courts sitting longer than ever with such decisions trying to not only decide the case in front of them but the potential long term effects of decisions they are making. Look here for the wide array of upcoming lawsuits related to AI. We don’t need a machine learning model to predict whether those will happen - just common sense.