Page 23 - Delaware Lawyer - Spring 2023

of the PAL program allowed for the enablement of voice-based interaction with civilian devices. Since Apple’s acquisition of the technology, other artificial intelligence-based personal assistants have come to market, such as Google Assistant and Amazon Alexa.14,15 ThisformofAIisintegrated into our everyday use.
AI is defined by Encyclopedia Britannica as “an area of computer science that deals with giving machines the ability to seem like they have human intelligence; i.e., the power of a machine to copy intelli- gent human behavior.”16
Artificial intelligence can be ana- lyzed and broken down in a myriad of ways. One way stipulates that there are two categories of AI — weak AI and strong AI. Weak AI or artificial narrow intelligence (ANI) is trained to perform specific tasks and is limited to a narrow area.17 Weak AI lacks hu- man consciousness; examples include Siri, IBM Watson and autonomous vehicles. Strong AI comprises both ar- tificial general intelligence (AGI) and artificial super intelligence (ASI). AGI, or general AI, is a “theoretical form of AI where a machine would have an intelligence equaled to humans; it would have a self-aware consciousness that has the ability to solve problems, learn and plan for the future.”18 ASI is known also as superintelligence and “would surpass the intelligence and ability of the human brain.”19
According to Arend Hintze, a re- search professor at Michigan State University, there are four modes of AI.20 The Reactive machines are task- specific computers that receive data and deliver consistent output — think website, vendor or movie recommenda- tions. The Limited Memory computer is programmed to observe patterns to act — consider self-driving cars.21 The Theory of Mind computer is a concept computer that combines both the
There are chatbots available that collect information from the internet and can use human voices to create conversations. The so-called large language models can create content that matches any musician's, speaker's or writer's speech patterns or writing styles.
reactive and limited memory com- puter.22 The Self-awareness computer model is another concept computer, which has people concerned about the replacement of human direction and decision making.23
When it comes to domestic vio- lence, the generative AI — a subgroup of the Reactive machine — receives data and produces outputs that appear to be human-created content.24 Chat- GPT, created by OpenAI, generates original written work based on data prompts.25 DALL-E2, also created by OpenAI, creates images, e.g. pic- tures based on written prompts.26 Other generative AI includes GitHub Copilot, which helps create code for users. There are chatbots available that collect information from the internet and can use human voices to create conversations.27 The so-called large language models (LLMs) can create content that matches any mu- sician’s, speaker’s or writer’s speech patterns or writing styles.28
Computer vision is the subfield of AI that teaches machines to under- stand images and videos. Deepfake or face swapping technology refers to the “specific kind of synthetic media where a person in an image or video
is swapped with another person's likeness.”29 It has become preva- lent, and continues to threaten the credibility of online information, especially visual media.
Results Are in the Hands of the User
According to Dr. Danny T. Barnes of Absolute Veritable Solutions, LLC, “AI is a great tool with great possi- bilities. Histor y has given humans a lot of great tools, but the heart of the user can make the tool prosperous or a pariah. AI is based on the data that it is fed. That is a danger with any false positive in data analysis. When using AI for legal matters, we must do everything to bring the number of false positives to zero, or legal can be the inertia to destroy innocent lives. One of the drawbacks of AI in the use of facial recognition technology is the disproportionate number of false positives for people of color, based on the technology’s default setting of ‘unrecognized,’ which translates to ‘threat’ or ‘hostile’ or a ‘concern.’”30
In the hands of bad actors, AI tools can manipulate victims by disseminat- ing disinformation through created content that highlights victims in a negative light. Domestic abusers can
SPRING 2023 DELAWARE LAWYER 21