apidays Helsinki & North 2023 - What Generative AI Really Means To Cloud Ecosystems, Merja Kajava, Aavista

apidays Helsinki & North 2023
API Ecosystems - Connecting Physical and Digital
June 5 & 6, 2023

What Generative AI Really Means To Cloud Ecosystems
Merja Kajava, CEO at Aavista

------

Check out our conferences at https://www.apidays.global/

Do you want to sponsor or talk at one of our conferences?
https://apidays.typeform.com/to/ILJeAaV8

Learn more on APIscene, the global media made by the community for the community:
https://www.apiscene.io

Explore the API ecosystem with the API Landscape:
https://apilandscape.apiscene.io/

Transcript

1. Aavista Oy Merja Kajava @mkajava APIDays Helsinki & North 2023 What Generative AI Really Means to Cloud Ecosystems
3. ChatGPT is the tip of the iceberg
4. OpenAI ecosystem
5. OpenAI ecosystem
6. AI is being embedded into tools Productivity apps Dev tools Search CRM Analytics Messaging ERP API and Plugins Copilots
7. AI in the cloud ecosystems
8. “We need to start using AI in our applications”
9. Where is the data coming from?
10. Data comes from several sources ERP Databases Documents … Internet 3rd party databases … Public Private Data provenance
11. Publish Your API with Plugins Copilot Your API Assistant Plugin for Google Plugin for ChatGPT Marketplace for ChatGPT
12. Plugin is the new API
13. Authenticate Plugin Users No authentication Service-level authentication User-specific authentication
14. Be responsible Identify risks Mind the privacy Consider IPR Validate usage via moderations Monitor the usage
15. From API to Prompt Development
16. Specify the input and include format for the response. PROMPT Getting Started with Prompt Development prompt = f""" Translate the following text to all the Nordic languages and provide the answer in the same language: 'What is the capital of Sweden?' Output in Markdown table with headers: Language, Translation, Answer """ response = get_completion(prompt) Text Files Images Speech
17. Response is produced and formatted according to the prompt. RESPONSE Responses to Prompts
18. Tuning the Model def get_completion(prompt, model="gpt-3.5-turbo"): messages = [{"role": "user", "content": prompt}] response = openai.ChatCompletion.create( model=model, messages=messages, temperature=0, ) return response.choices[0].message["content"] TEMPERATURE Decide the degree of randomness of the model's output FOUNDATION MODEL Choose the model, for example GPT 3.5 Turbo, or GPT 4.
19. Pricing models are different Words from the prompt and response are the basis for the costs. Foundation models also have different pricing. OpenAI API Prices are per 1,000 tokens. Token is a piece of word, where 1,000 tokens is about 750 words. Generative AI on Vertex AI Prices are per 1,000 characters. Characters are counted by UTF-8 code points.
20. Be creative Start experimenting Build plugins on top of your APIs Compare prices Remember the Responsible AI
21. The Data Refinery Company Merja Kajava https://www.linkedin.com/in/merjakajava