Artificial Intelligence : 4 Key Challenges Facing Engineers

Many of the biggest challenges that people working with artificial intelligence (AI) will face will be nontechnical. AI is often inappropriately treated as a magic bullet for solving ALL of the world’s problems or, at the very least, keeping businesses on the competitive edge. The internet is filled with stories of AI that resulted in negative human impact and unintended consequences. If you are developing or plan to develop artificial intelligence solutions, here are some questions for your consideration:

  1. Lack of clarity, transparency, and alignment regarding WHO and WHAT is most important.
  • What are your success metrics?
  • What does failure look like?
  • Does your vision of failure address who/what should not be critically impacted by your technology?
  • Does the impact of your actions align with your stated intention?

2. More data, less insight.

  • How do you determine what’s true and when it’s true?
  • Beyond data dictionaries and schemas, who provides the environmental and historical context for any datasets used on your project?
  • What was true at the time the data was collected? What is true now?
  • Does your use of the data consider historical context?
  • Should you be using this data? What would you do if you could NOT use this data?
  • Who will address the potential evolution of the datasets that you’ll manage and grow?
  • Under what situation should your data be used outside of the scope of this technology?

3. Misuses and abuses.

  • In general, where is AI helpful?
  • How is AI useful for your project?
  • Can your solution exist without AI?
  • In what environments is your technology not suitable? Why?
  • How does your technology disrupt any existing sectors, humans or environments?
  • Define the “golden path” for using your technology (ideal user personas, ideal workflows, ideal environments etc.). Does your technology allow alternative paths of use?

4. Siloed vision.

  • As it relates to your technology, who should shape its future?
  • Who are the people that will be positively impacted by this technology?
  • Who are the people that may be negatively affected by this technology?
  • Are they key members of the team, and do they have any real influence on this technology’s vision?

Total
5
Shares
Related Posts
Read More

Using Pseudocode to Solve Complex Problems

Pseudocode is not just for novice programmers, It is an excellent problem solving tool, particularly for writing complex algorithms. It also works well as a follow-up to creating well thought-out user scenarios and user stories. It is a great method for uncovering unclear decisions, hidden side effects, and for defining all inputs, outputs and interactions needed to effectively solve a problem.