Slide 1

Slide 1 text

AI for leaders: from strategy to practice Nikolay Stoitsev Senior Engineering Manager at StorPool Storage

Slide 2

Slide 2 text

No content

Slide 3

Slide 3 text

Caution, optimism, and skepticism

Slide 4

Slide 4 text

Privacy Inadvertent sharing of secrets

Slide 5

Slide 5 text

Accuracy Hallucinations or stale data

Slide 6

Slide 6 text

Intellectual property infringement Violating IP, copyright or trademark

Slide 7

Slide 7 text

You can’t ban AI

Slide 8

Slide 8 text

You lose competitive advantage

Slide 9

Slide 9 text

People will still use it

Slide 10

Slide 10 text

LLMs can’t think

Slide 11

Slide 11 text

The intelligence illusion is in the mind of the user

Slide 12

Slide 12 text

Deepseek R1 “While LLMs demonstrate behaviors that resemble reasoning in specific contexts, their capabilities differ fundamentally from human cognition”.

Slide 13

Slide 13 text

No content

Slide 14

Slide 14 text

Crunching PDFs Demo

Slide 15

Slide 15 text

• ChatGPT • Gemini • Grok • DeepSeek • Perplexity Web search

Slide 16

Slide 16 text

Can provide feedback and different perspective Demo

Slide 17

Slide 17 text

No content

Slide 18

Slide 18 text

Tasks that require quantity Demo

Slide 19

Slide 19 text

Tasks where you are an expert

Slide 20

Slide 20 text

Tasks to transform information

Slide 21

Slide 21 text

Tasks for data analysis Demo

Slide 22

Slide 22 text

Tasks that stretch you over many disciplines

Slide 23

Slide 23 text

Brainstorming with Live Mode Demo

Slide 24

Slide 24 text

No content

Slide 25

Slide 25 text

• Team needs • Business value • Technical feasibility Where to start?

Slide 26

Slide 26 text

• Vendor’s expertise • Privacy/security measures • Integration with existing work fl ows How to pick a tool?

Slide 27

Slide 27 text

Each team has specific needs

Slide 28

Slide 28 text

Help them understand testing

Slide 29

Slide 29 text

Think about the end-to-end workflow

Slide 30

Slide 30 text

No content

Slide 31

Slide 31 text

• Hallucinations • IP indemnity • Unknown failure modes Remind about limitations

Slide 32

Slide 32 text

Lead by example Use AI daily and share your experience

Slide 33

Slide 33 text

• What tools are allowed • What information should and should not go in • Verifying the output • Standard procedure Set the rules

Slide 34

Slide 34 text

Need to measure the RoI / efficiency gains

Slide 35

Slide 35 text

Managing degrading quality https://www.gitclear.com/ coding_on_copilot_data_shows_ais_down ward_pressure_on_code_quality

Slide 36

Slide 36 text

With more AI we need more managers Alekseeva, Liudmila and Azar, José and Gine, Mireia and Samila, Sampsa, AI Adoption and the Demand for Managerial Expertise (September 27, 2024). Available at SSRN: https://ssrn.com/abstract=4986379

Slide 37

Slide 37 text

Q&A Nikolay Stoitsev https://www.linkedin.com/in/nikolaystoitsev/