Özhan Atalı

My AI notes - 3/3

31 Eki 2024 - İstanbul

Building AI in your company

Itdoes take maybe two or three years for a company to become good at AI.

Case study: Building a Smart speaker

AI Pipeline to Processes for command: 1-Trigger word/Wakeword detection :Hey device 2-Speech recognition :Recognize the audio 3-Intent recognition :Tell me a joke/time/music 4-Execute joke

Self-driving cars: input A:Image/Radar/Lidar/GPS

  1. Car detection:supervised learning
  2. Pedestrian detection:supervised learning
  3. Lane detection
  4. Traffic light detection
  5. Other obstacles than cars
  6. Motion planning:steer/accelerate/brake
  7. Trajectory prediction

Roles of an AI team

AI Transformation Playbook

  1. Executing pilot projects to gain momentum (6-12 months. can be inhouse/outsource)
  2. Build an in-house AI team > to support other seperate Business Units/Lines (matrix hierarchy)
  3. AI team under a senior manager of tech/innovation/it etc. CEO to support/funding
  4. Broad AI training
    • Executives/Senior What can AI do » /AI strategy/Resource allocation
    • Leaders » Set project direction/reousece allocation/monitor process
    • AI engineer trainees » Build and ship AI projects/gather data/execute on AI projects
  5. Develope AI strategy (not the step1)
  6. Develop internal/external relations
    • Invester relations
    • Government relations
    • Customer/user education
    • Talent/recruiting
    • Internal communications

AI Strategy

AI pitfalls to avoid

Applications

Goldilocks rule for AI

Social Impacts








📝 | Sitenin son güncellenme tarihi: 08 Kas 2024