The Black Box of Artificial Intelligence

June 20, 2023
8:06 pm

We are entering a new age of the industrial revolution. First with the invention of steam engines that paved the way to increased production, second with the discovery of electricity with advances not only in manufacturing but also in communication, third with the introduction of computers, which led to progressive automation, and finally with the emergence of artificial intelligence technology, we are heralded toward the fourth epoch in the industrial revolution. It impacts trade, labor market organization, and information access. It affects politics, economics, and culture. It affects those who directly augment it and those who indirectly benefit from its use and advancement.

Artificial Intelligence also involves how we think, remember, and reason individually and with others. It greatly improves our way of life, but due to its ubiquity and inherent complexity, like all technological advancements, there will be pushbacks and resistance. While some imagined fears that are rooted in sci-fi literature, others took advantage and positioned themselves early in the game. It is highly complex and enigmatic to many. That is why there is frequently a particular concern regarding its transparency. This assumes that members of a profession and the profession as a whole maintain a heavily guarded black box safely tucked away from the prying eyes of the public. It is also assumed that these individuals have a good grip on their actions and can therefore control them, or at least insure against an unexpected loss of control. Expertise should require explanation. Honesty and transparency necessitate explaining acts and motives. But how can we determine whether an algorithm is operating ethically if not even its creators know precisely how it operates?

Answering such difficult concerns may require considering agency and autonomy in humans and machines. Transparency in AI systems means stakeholders sharing data, procedures, uses, and outputs. It is how an organization communicates with stakeholders about the system’s components and functions. This empowers executives, managers, and consumers to make informed decisions on AI technologies in their roles. Information quality and stakeholder comprehension are factors.

When a system is in place that lacks transparency, how can we ensure ethical behavior? Answer: visibility—how people perceive you. That means visibility for others. You’ll behave better if others can see you. The likelihood of good behavior increases when others can observe it. Good communication and participation in public conversations and consultations are virtues. ASTN Group initially published a whitepaper titled “Intelligent Agents for Information Augmentation Using Novel Approaches to Data Analysis” to promote transparency and accountability. In it, the approaches were discussed and defined. Core ethical principles are embedded in Valmiz. Additionally, the lead developer has carefully defined the scope and capabilities of Valmiz, which is not attempting to solve every issue known to man. ASTN Group understands that explaining every aspect of Valmiz and how it works is of paramount importance. What problems Valmiz is trying to solve and how it can be a solution. Moreover, ASTN Group also provided an executive summary for investors presenting the future development and trajectory of Valmiz and what target markets and industries Valmiz can be part of.

ASTN Group also actively participates in forums where we can discuss relevant issues, like the upcoming AI Asia Expo 2023 in Manila. Here, innovators and experts from the global AI community will converge to foster dialog that will set the road map for the AI industry in the Philippines. Talks between agencies establish accountability in the community, which follows revision and critique. There must be code revision, whistleblowing protocols, and excellent communication to reduce the necessity for whistleblowing. This level of peer examination is desirable and necessary. There are frequent concerns that extremely powerful corporations armed with the inscrutable mechanisms of artificial intelligence could manipulate people without their knowledge. But with involvement and visibility, this is mitigated.

How much transparency is acceptable for some activities is unclear. Commercial companies also need privacy. Given serious opponents, only a fool wants the security services of their company to be entirely transparent, but determining the line may be challenging. How about some competitive advantages that must remain behind closed doors? Transparency is not either all or nothing. How much information to share depends on how the AI will affect stakeholders. Here are some steps to look into to answer the question at hand: First, consider the complete scope of parties involved in the system and the level of disclosure required to gain everyone’s faith in AI. Stakeholders within an organization need to be afforded certain levels of openness in order to effectively contribute ideas and perspectives. And second, establish methods for gathering, documenting, and reporting information that may be packaged for internal and external users after establishing the levels of openness required for each stakeholder. Transparency in AI emerges throughout the lifecycle. ASTN Group will continue its commitment to be transparent, and by fostering communication and cooperation between departments, transparency helps teams solve challenges and generate new ideas.

June 20, 2023
8:06 pm