What is artificial intelligence (AI)? 

The goal of artificial intelligence (AI), sometimes called machine learning (ML), is the simulation of human intelligence by a computer. Traditionally, computers were programmed to make decisions using conditional logic. We can use the example of self-driving cars: if an animal is in the road, then swerve to avoid it. However, conditional logic cannot take into account every scenario nor act with human-like intelligence derived from learning, reasoning, and perception. With AI, the computer will weigh the consequences of its actions, like humans would.

What makes a successful AI/ML model?  

Computers are taught in a similar way to humans. Teaching a computer has at least these 3 components:

  1. The architect (usually a human) – creator of the AI and designs the training lessons
  2. The instructor (usually a combination of human and computer) – teaches the AI
  3. The learner (the computer) – the AI learning the lessons

To successfully implement AI, these boil down to 4 components:

  1. Trusted data source/lake for training (and using a training platform such as Labelbox)
  2. Choosing the computer algorithm for the intended project
  3. Design and build an infrastructure (on-premises and cloud resources)
  4. Constantly test, validate, monitor, and retrain the model

When is AI the solution?  

When there is a problem with many different solutions but need to find the optimal solution. This is called an optimization problem.

Example #1:

Aviation – Airlines needs to make sure their crew is always efficiently scheduled and paired for each of their routes.

Large datasets and want to find patterns that would be too difficult for anyone or team of humans.

Example #2:

Healthcare  -Finding drug interactions to make drug prescriptions safer.

Recommend treatment based on symptoms.

Military – Monitoring satellite images looking for anomalies.

How can GovSmart support a customer’s AI mission?  

  • Hybrid on-premises/cloud data storage (NetApp)
  • On-premises/cloud AI data trusted data training (Labelbox)
  • On-premises computing (HPE, Dell EMC, Red Hat) & networking (Cisco, f5)
  • Monitoring solutions (Solarwinds, Splunk, AppDynamics)

 

If you have any questions about this or other topics, please do not hesitate to reach out to us!