ディープオートメーション：大規模な製造プロジェクトで学んだCIOの教訓

当レポートは、ディープオートメーションに焦点を当て、市場の概要と技術バイヤーへのアドバイスなどについてまとめています。

目次

エグゼクティブスナップショット

状況概要

• 機械に「知」を教える
• 光の速さで意思決定を自動化
• モデルをビジネスコンパニオンと強力なUIに変換する
• ディープオートメーション管理の課題への対処

技術バイヤーへのアドバイス

• 中小企業の深い知識を必要とする、小規模で高度に焦点を絞ったディープオートメーションプロジェクトを設計する
• ビジネス仮説をサポートする知識と因果関係を捉える
• モデルUIを設計して、生産性の高いビジネスコンパニオンとして機能するようにする
• 変更管理計画を設計して、因果モデルが競争上の差別化要因になる方法を決定する

参考資料

This IDC Perspective provides insights into a new dimension of automation that is emerging: deep automation. With deep automation, you can automate almost anything that is highly "teachable," using language. That means AI can be applied to almost any manufacturing or business process.Deep automation captures expert knowledge and stores it in a graph database using structural causal model representations. This becomes a collective intelligence knowledge base that can be continuously "taught" using causal AI to expose cause-and-effect relationships, which become the basis for automated unbiased business decisions.While most AI/machine learning solutions require (and are limited by) a massive source of machine-generated data, the unique knowledge base for deep automation is taught to the model from readily available language-based documents including books, white papers, and subject matter expert (SME) interviews.To move from teaching knowledge to the model to having it make automated decisions requires a newly rediscovered technology: the structural causal model technology first developed by Judea Pearl in the 1980s. This model enables an organization to use its unique knowledge base to massively expand its decision-making capabilities by exploiting the model's capability to navigate through impedance gates with minimum friction.While users and subject matter experts will teach the model their best practices, the quality of the output from the model, the UI, will determine how efficient and effective the machine-human symbiosis is; in essence, it becomes a "business companion" to the user."Just the idea of deep automation, teaching machines to make complex decisions, can appear to be a daunting challenge to an organization, from both the business perspective and the IT perspective," says Robert Multhaup, adjunct research advisor with IDC's IT Executive Programs (IEP). "The CIO sits in the middle of this dilemma and is in a unique position to steer the organization through this labyrinth of new ways of doing business, a new human-machine intersection, a new culture of knowledge sharing, and a new competitive weapon."

Executive Snapshot

Situation Overview

• Teaching "Knowledge" to Machines
• Automating Decision Making at the Speed of Light
• Transforming the Model into a Business Companion and a Strong UI
• Addressing Deep Automation Management Challenges

Advice for the Technology Buyer

• Design a Small, Highly Focused Deep Automation Project Requiring Deep SME Knowledge
• Capture the Knowledge and Cause-and-Effect Relationships to Support the Business Hypothesis
• Design the Model UI to Ensure It Operates as a Highly Productive Business Companion
• Design a Change Management Plan to Determine How Causal Models Can Become a Competitive Differentiator

Learn More

• Related Research
• Synopsis