Skip to content

Latest commit

 

History

History
262 lines (175 loc) · 3 KB

slides.md

File metadata and controls

262 lines (175 loc) · 3 KB
marp theme
true
uncover

bg fit

How (not) to (use) AI

What are we talking about

$ whoami
Wtf is AI
Machine vision
Larry
Generative AI
Unsupervised AI
Manufacturing
Hardware & Platform
Take it to the cloud

tleilaxu@axians:~$ whoami
Name: Kenneth Van den Driessche
Work: Axians
Time: 5y1m1d19h
Current job: Consultant@Pfizer

Previously: Social media platform developement @ travel sector Development of backoffice software and network architecture @ event sector Create a more data-driven organisation @ shipping

Be customer oriented Be team fix it - deadlines not optional Data is valuable

Smart factory technology Innovation Technical consultant

(Whatever that means)

Machine vision

Deep learning vs traditional Computer vision

bg auto

bg auto

bg auto bg auto

Pose estimation

PPE detection

bg fit

Adversial AI

Generative AI

bg fit

bg fit

bg fit

Larry

Cognitive AI

I herd you like AI so we put some AI in AI so you can AI while you AI

bg fit

bg fit

bg fit

bg

Applications

Talk to production line

Expected

Prompt:
What whas the avrage temperature for the current production item
Answer:
15C

Early gpt example

Prompt:
Please find the lowest Amount in the table below.
| Description | Amount |
| Pants       | 5      |
| Skirt       | 5      |
| Shoes       | 5      |
-------------------------------------------------
Response:
| Panda       | 5      |
| Tie         | 5      |
| Glasses     | 5      |

Generative AI will never be 100% reliable.

Generate a event log, and use AI to scroll to the right position.

How not to AI

bg fit

bg fit

Data FIRST

Why not generate the data for you model

Bad data Rigid pipeline Framented data

bg fit

Questions?

bg fit

Next up: 05/12/2023

Open space

bg fit

https://github.com/tl3ilaxu/aitalk