-
1주차
-
알고리즘 vs 인공지능
- 알고리즘
- input → computer → output
- 인공지능
- input/output → computer → program
- 프로그램으로 만들 수는 있음
- // 알고리즘, 인공지능, 컴퓨터 차이
- 모델링
- 알고리즘
-
인공지능과 프로그램을 구분하지 않으면 안됨
-
sequence of instruction
- 컴퓨터와 계산기의 차이
- 더 명확안 컴퓨터의 정의는 turing machine이다
-
dead internet theory
- 중앙화된 플랫폼이 이걸 막고 있음 (ai 도배)
- 아직은 틈이 있지만, 나중에는 구별이 안될것임
- what happens to the billion other websites that needs traffic?
- people will just use ai tools, with no need to visit websites
- then the websites will have less incentive to create websites
- so we need a new protocol
- so we need to learn the A2A protocol
- incentive alignment
- “인공지능의 진짜 의미” <<< 영상 찾아보기
-
claude made the protocol
- try chatgpt cmp
-
mcp 여러개를 동시에 agent에 붙히면?
- 문제점
- 검색의 문제
- mcp하나 잘 만드면 무슨 api를 호출할지 아는데 mcp여러개를 만드면 헷갈림
- 같은 계열의 mcp를 모으기
- 문제점
-
google
- gemini와 그 외 모든 인공지능 ai agent와 여러개의 a2a
- 구글 클라우드안에 있는데, 클라우드쪽으로는 aws에게 지고 있음
- 기업 내부에서 에이전트를 하는걸로
- 지금 테스트기간
- 구글의 a2a 찾아보기 <<<
- openai vs gemini
- gemini has more accurate real results (ex. stock value today) → chatgpt has outdated results
- 데이퍼 provider과의 계약을 얼마나 하는가
- gemini has more accurate real results (ex. stock value today) → chatgpt has outdated results
-
agent card
- name, description, skills
-
mcp vs a2a protocol