The IBM website lists the following benefits and risks of AI

AI Benefits

• Automation of repetitive tasks.
• More and faster insight from data.
• Enhanced decision-making.
• Fewer human errors.
• 24×7 availability.
• Reduced physical risks.

AI challenges and risks

Organisations are scrambling to take advantage of the latest AI technologies and capitalise on AI’s many benefits. This rapid adoption is necessary, but adopting and maintaining AI workflows comes with challenges and risks.  Companies involved with developing AI will continue to downplay the dangers of AI except in cases where the marketing department can use those dangers to ‘Hype’ their product. Notable absences from the risks below are the issues of Data Centres and their effects on the environment, and the effects on AI on Employment.

• Data risks – AI systems rely on data sets that might be vulnerable to data poisoning, data tampering, data bias or cyberattacks that can lead to data breaches. Organisations can mitigate these risks by protecting data integrity and implementing security and availability throughout the entire AI lifecycle, from development to training and deployment and post-deployment.
• Model risks – Threat actors (individuals or groups that intentionally cause harm to digital devices or systems.) can target AI models for theft, reverse engineering or unauthorised manipulation. Attackers might compromise a model’s integrity by tampering with its architecture, weights or parameters; the core components that determine a model’s behaviour, accuracy and performance.
• Operational risks – Like all technologies, models are susceptible to operational risks such as model drift, bias and breakdowns in the governance structure. Left unaddressed, these risks can lead to system failures and cybersecurity vulnerabilities that threat actors can use.

The challenges and risks above are from the perspective of AI companies and developers which means this is not a complete list.  There are other dangers to the economy, society and our civilisation as a whole. The most serious risk is the existential risk to humans.  This is the subject of heated debates and discussions.  Opponents of AI argue that Western economies (especially the US) should slow-down AI development until we have a better understanding of its effects (positive and negative).  Proponents of AI argue that western economies (especially the US) must be competitive with China when it comes to AI. They believe that China poses a greater threat to western civilisation than AI.

 

AI alignment

AI alignment aims to make AI systems behave in line with human intentions and values