How do you evaluate the ROI of an investment in generative AI?

With great difficulty:

Unless you're actively tracking your time, and the time of your employees it's hard to get efficiency improvement metrics.

With this in mind, you might want to take a softer approach:

1/ Are our team members HAPPIER that they can spend time on more fulfilling work
2/ Are you able to reduce the amount of REPEAT tasks you're carrying out?
3/ Are the more mundane activities QUICKER to complete with Gen AI?
4/ Are you surfacing any data INSIGHT that you couldn't before?

Given the current status of AI I believe that organisations should create use cases for what they believe can be achieved.  Trying to then apply a ROI at this stage I think is futile as it will be so vague and full of assumptions to render it worthless.  Organisations should recognise that investment is a good thing and will deliver benefits if done correctly but evaluating it is just spurious accuracy at the moment. 

The evaluation of the return on investment (ROI) for generative AI involves assessing both the potential benefits and costs associated with the technology. It is pertinent to note that the evaluation of ROI for generative AI investments can be complex, as it involves both tangible and intangible factors. Additionally, the accuracy of ROI calculations depends on the accuracy of your benefit and cost estimates. In some cases, it might be beneficial to engage financial experts who have experience in evaluating technology investments to ensure a thorough analysis.

ROI of investment in GEN AI depends upon long term benefits derived by cost of implementation versus manhours saved. The accuracy and usability will also play an important role. The ROI is not only quantitative but will also be qualitative.