Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Using AI in Software Design: How ChatGPT Can Help With Creating a Solution Architecture

Using AI in Software Design: How ChatGPT Can Help With Creating a Solution Architecture

Artificial intelligence, or AI, is becoming a big part of our lives. One type of AI is Large Language Models (LLMs), like chatGPT. At the iSAQB Software Architecture Gathering, I showed how chatGPT can help with complex tasks in software design, such as the tasks in the iSAQB Advanced Exam.

We delved into the intricacies of conversing with chatGPT, illustrating the strategies needed to generate productive prompts and effectively utilize the model as a sparring partner. The talk exemplified these concepts by taking the audience on a step-by-step journey through the process of tackling the iSAQB Advanced Example Exam, utilizing chatGPT.

We demonstrated how to prepare for a chat session with chatGPT, how to generate robust prompts, and how to manage the chat for optimal outcomes. We aimed to highlight the model’s capabilities as an interactive tool that can provide valuable insights and streamline the process of developing software architecture.

Whether you are a seasoned architect or a novice in the field, this presentation will equip you with novel techniques to navigate the challenging landscape of software architecture with the assistance of AI.

Ralf D. Müller

November 28, 2023
Tweet

More Decks by Ralf D. Müller

Other Decks in Programming

Transcript

  1. [email protected]
    @ralfdmueller



    https://www.linkedin.com/in/rdmueller/

    View full-size slide

  2. GPT-3 (Generative Pre-trained Transformer 3) was released in 2020. It was pre-
    trained on a massive dataset of 570GB of text data and had a capacity of 175
    billion parameters. It was fine-tuned for a wide range of language tasks, such as
    text generation, language translation, and question answering.
    GPT-4 (Generative Pre-trained Transformer 4) was released in 2021, it was pre-
    trained on a massive dataset of many terabytes of text data and had a capacity
    of over 500 billion parameters. It was fine-tuned for a wide range of language
    tasks, such as text generation, language translation, and question answering
    with even more accuracy and fluency than GPT-3.

    View full-size slide

  3. Comparative performance of humans versus GPT-4.0 and GPT-3.5 in the self-
    assessment program of American Academy of Ophthalmology
    Out of 1023 questions,
    GPT-4.0 achieved the best score (82.4%),
    followed by humans (75.7%)
    and GPT-3.5 (65.9%)
    *Ophthalmologie ist der Zweig der Medizin, der sich mit der
    Anatomie, Physiologie und den Krankheiten des Auges
    beschäftigt.

    View full-size slide

  4. GPT-3: 2.048 Token
    GPT-4: 32.000 Token
    (etwa 25,000 Words)

    View full-size slide

  5. I am an experienced software architect in the field of web
    development. My favourite programming language is Java or Groovy.
    JavaScript and corresponding frameworks are not my thing. I prefer
    to get by with minimal JavaScript.

    View full-size slide

  6. I am an experienced software architect in the field
    of web development. My favourite programming
    language is Java or Groovy. JavaScript and
    corresponding frameworks are not my thing. I
    prefer to get by with minimal JavaScript.
    I am an experienced software architect in the field of web
    development. My favourite programming language is
    Python and JavaScript. I love modern JavaScript
    frameworks.

    View full-size slide

  7. I am an experienced software architect in the field
    of web development. My favourite programming
    language is Java or Groovy. JavaScript and
    corresponding frameworks are not my thing. I
    prefer to get by with minimal JavaScript.
    My expertise extends to modern technologies and approaches
    such as Spring Boot for the creation of robust and scalable backend
    services. I appreciate the lightweight and responsive design of
    Twitter Bootstrap, which I use specifically to create efficient user
    interfaces. Documentation is an integral part of my work, favouring
    the docs-as-code approach as it allows documentation to be
    maintained as part of the development code, ensuring timeliness
    and quality. These tools and methodologies form the foundation of
    my work, allow me to have an agile and responsive development
    environment, and help me to write clean, maintainable code.

    View full-size slide

  8. What would you like ChatGPT to know about you to provide better responses?

    View full-size slide

  9. https://twitter.com/dr_cintas/status/1691160385297006593

    View full-size slide

  10. https://twitter.com/dr_cintas/status/1691160385297006593

    View full-size slide





  11. View full-size slide

  12. Thanx for your attention!

    View full-size slide