Benefits attributable to incorporating AI in companies
While the benefits attributable to incorporating AI in companies are endless, there has emerged a new type of risk for companies already utilizing machine-based decision systems. Aside from the social risk of bias and the financial risks that could make modeling errors, companies have to account for IT security risks, litigation risk, reputational risks, and regulatory risks that might occur due to AI-based decisions. Managers will soon be held accountable for machine-based decision failures. Nonetheless, the rollout of AI cannot be halted.
Cybersecurity raises both alerts and provides a solution against AI-based failures. A majority of firms perform cybersecurity audits with the responsibility and liability, having a broad reach even on the board of directors. Companies using AI models for socially and consequential decisions need to introduce similar audits. Based on the proposed Algorithmic Accountability Act, large companies might soon be required to formally evaluate their high-risk automated decision systems for accuracy and fairness. Consumers will have the right to an explanation when companies use algorithms to make automated decisions. The Information Commissioner’s Office (ICO) of the UK recently accepted comments for its proposed AI auditing framework, which will be more extensive.
This framework will facilitate ICO’s compliance assessments of companies utilizing AI automated decisions. Among the risk areas identified using the framework include fairness and transparency, accuracy and security and governance, and accountability XX(https://knowledge.wharton.upenn.edu/article/audits-way-forward-ai-governance/). ICO is still furthering its plans, but it might take a while before a regulatory audit framework comes into law. Forward-thinking companies are, however, not waiting for regulation. Massive AI failures will diminish customer trust and cause for stricter regulations. Such ought to be avoided today through proactive measures.
The audit process will commence with tracking the entire inventory of all machine learning models in use within the company, along with the specific applications of such models. Developer and business owner model names and risk rating will also be assessed. In case the model audit is approved, it would evaluate the inputs, model, and outputs of the data (https://knowledge.wharton.upenn.edu/article/audits-way-forward-ai-governance/)xx. the training data will also need to be evaluated for data quality and possible biases concealed within the data.
The Knowlwge@Wharton facilitates coverage of Wharton facility research. It uses a navy blue color, which makes it popular for both sexes visiting the site. It is user friendly and easy to navigate through the utilization of four drops down menus for various categories. However, the dropdown menus do not reveal clear submenus. The content on the homepage utilizes a balance between images and words. The site has enticing calls to actions designed in such a way that hovering over the images reveals additional information on the content of the article. Other call to action includes a “Listen on iTunes” button well positioned on the homepage and login and register links on the top of the webpage. The stay informed bar with an email address submission bar and search bar both provide for convenient capture. The site also employs a professional tone and with a site language translation feature enabling its to cater for the diversified users across the globe who may not understand how to read in English. Social media integration has been adequately performed on the site for Facebook, Instagram, Twitter, and LinkedIn platforms. The site is, however, missing a live chat feature and should consider incorporating it on the next update.