Artificial Intelligence is undoubtedly the most revolutionary technology of the 21st century. There is hardly a month that goes by without the announcement of a new AI product or trend. McKinsey estimates AI to add an additional $13tn to the global economic output by 2030.
Businesses are increasingly looking for ways to leverage AI to boost productivity, efficiency, customer experience, profitability, and business results. While there are several emerging use cases and business benefits of AI, there are also certain risks to keep in mind while adopting it.
Enterprise leaders are finding that along with significant benefits, AI also brings in unique and unprecedented risks that need to be addressed. Just like any new technology, AI is a double-edged sword – but the edges are far sharper if not fully understood.
Let’s explore some of the biggest and most common AI risks:
1) Privacy and Compliance
Organizations today collect, integrate and harness humongous amounts of data from various sources. And AI needs this data to deliver accurate findings and recommendations. However, this data can also contain sensitive information related to customers and PII. Inadvertent usage of this data for AI algorithms can lead to non compliance risks related to General Data Protection Regulation (GDPR) or the California Consumer Privacy Act (CCPA).
The solution to mitigating this risk can be by using AI itself. Gartner predicts that over 40% of privacy compliance technology will rely on artificial intelligence in the next three years. AI can be used to automate the manual and burdensome approach to data discovery and SRR management.
Learn More: Illuminate Your Dark Data
2) Discrimination and Bias
AI is not independent of biases. Since AI algorithms are built by humans, human biases can be introduced, intentionally or inadvertently, into the AI algorithm. Moreover, the machine learning capabilities of the AI system can also make it inherit biases. These biases will produce results that favor certain aspects over others, which will damage the reputation and trust in the company. Therefore, businesses must build the ethical framework into the AI models and audit the data regularly with the goal to identify inconsistencies.
3) Employee Adoption and Trust
Employee adoption is critical to the success of any AI initiative. If companies don’t clearly communicate the objectives of an AI program and win employees’ trust, the entire project might fail. You could deploy the best AI solution but if your employees don’t trust and adopt it, the ROI suffers. Many workers are already worried about losing their jobs to AI softwares.
Leaders should strive to communicate on what AI can and cannot do. Debunk common AI myths among your workforce. Educate and train them on the software’s capabilities.
Enable a two-way communication and address their concerns. Leverage a Modern Intranet to improve communication and collaboration and build an AI-driven culture.
4) Security
Fraudsters and bad actors are increasingly trying to steal and exploit sensitive customer, marketing and financial data that companies collect for their AI initiatives. A lack of sufficient security precautions results in the creation of false identities. Organizations could still face customer backlash and regulatory challenges even though they are unwitting accomplices.
Some steps to mitigate this risk could be:
- Set up real-time monitoring of capabilities that identify and respond to new threats and vulnerabilities
- Limit data and model access for employees, contractors and third parties to only what’s necessary
Examples of Risks With AI
1. Royal Free breached UK data law in 1.6m patient deal with Google's DeepMind
London’s Royal Free hospital failed to comply with the UK’s Data Protection Act after it used 1.6 million patients’ data for its AI project with Google’s DeepMind Technologies without obtaining proper patient consent. This resulted in the organization being asked to commission a third-party audit, undergo a privacy assessment, demonstrate how it will comply with its duties in the future and establish a thorough legal basis for its AI project.
2. Facebook shuts down Admiral Insurance’s AI app due to user data privacy violation
Facebook has blocked the AI app (firstcarquote) of UK’s leading insurance company, Admiral Insurance, for exploiting the social media platform’s data to set insurance premiums. The app targeted first-time car owners and offered to analyze their Facebook data to check if they could be successful drivers. Users were offered savings of over $350 per year on car insurance if they are identified to be well-organized.
3. MD Anderson Benches IBM’s “Watson for Oncology” after unsafe treatment recommendations
After investing more than $62 million in its AI project with IBM’s “Watson for Oncology”, MD Anderson Cancer Center shelved the initiative. IBM’s role was to provide oncologists with valuable insights from the cancer center’s patient and research database. However, it was later found that the AI software was providing erroneous treatment suggestions due to inaccurate training data. Instead of using real patient data, IBM’s engineers used data of a small number of hypothetical cancer patients.
4. Amazon shuts down AI recruiting tool that showed gender bias
In a bid to fill thousands of job vacancies, speed up candidate hiring process and automate recruitment, Amazon deployed an AI recruiting software. But it was later identified that the algorithm tended to favor males over females. Everything ended up in a PR fiasco as the algorithm favored candidates who tend to use masculine language in their resumes.
Conclusion
Since AI is relatively new, few business leaders have had the opportunity to understand the full scope of its risks. They may overlook potential perils or overestimate certain risk-mitigation capabilities. Therefore, AI’s impact will depend not on the technology itself, but on how quickly, broadly, and wisely businesses can implement the necessary reforms.
If you’d like to learn more about this topic, please feel free to get in touch with one of our IPA consultants for a personalized consultation. You might also be interested in exploring our process automation solutions and services for further insights!
