Data and analytic strategies for developing ethical it
2 likes238 views
The document discusses strategies for developing ethical IT practices, highlighting the current challenges and issues in IT, including the misalignment of technology and business goals, big data misuse, and the opaque nature of machine learning algorithms. Recommendations include fostering IT-business partnerships, focusing on purposeful data collection, and encouraging education in algorithms and statistics to improve ethical standards. It emphasizes the need for organizations to adopt a long-term, sustainable approach to technology and data management to mitigate fragility and unethical practices.
Data and analytic strategies for developing ethical it
- 1. DATA AND ANALYTICS STRATEGIES FOR DEVELOPING ETHICAL IT Hyoun Park Founder and CEO, Amalgam Insights
- 2. AGENDA • About Us • Your Challenge for Ethical IT • Why IT is Broken • Identifying IT Issues • Recommended Vendors • Recommended Strategies
- 3. ABOUT AMALGAM INSIGHTS • Advisory and Consulting Firm focused on Technology Consumption Management • Bridging CFO and CIO challenges for managing technology at scale. • Tactically includes • IT Finance • Machine Learning Prep • Accounting and Audit Automation • DevOps • Enterprise Collaboration • Training and Retention
- 4. HYOUN PARK Founder and Enterprise Tech Whisperer, Amalgam Insights, an advisory firm focused on Technology Consumption Management Previously: Chief Research Officer at Blue Hill Research IT Manager at Bose, Teradyne Campaign Treasurer, Boston City Councilor Fantasy Baseball Prognosticator
- 6. FIND JUST ONE IDEA FROM THIS PRESENTATION THAT YOU CAN TAKE BACK INTO YOUR ENTERPRISE
- 7. IDENTIFY THE ENTERPRISE ARCHITECT, CTO, CIO, DATA, ANALYTICS, OR DATA SCIENCE EXECUTIVE WHO CAN WORK ON THIS
- 8. USE THIS INFORMATION TO BECOME A MORE ETHICAL IT ORGANIZATION
- 9. IT IS BROKEN
- 10. In the beginning Now Nature of IT Work Skilled Work Combo of outsourcing, “cloud,” traditional IT, science, and magic Definition of IT Work Certification, Governed and consistent standards Tech moves too fast for certificates Goal of IT Work Focus on building and supporting tech infrastructure Bring Your Own Everything, patchwork security, reactive governance Funding for IT Work Big Budgets Shrinking Budgets
- 11. THESE CHANGES HAVE ALSO MADE IT LESS ETHICAL
- 12. Is it the Tech? Tech allows us to do things more quickly, but not inherently evil Is it the People? Somewhat, but more due to inaction rather than active evil Is it the Process? Yes. IT has adopted a variety of unethical practices.
- 13. ACROSS DATA, APPS, ANALYTICS, MACHINE LEARNING, AND IT SUPPORT, WE HAVE BEEN TRAINED TO DO THE WRONG THING
- 14. BUT WE CAN FIX IT
- 15. BREAKING OUT THE PROBLEMS • Traditional Technology and Do-It-Yourself IT • IT-Business Alignment defined as Outcomes-Based efforts • Big Data as an Opt-Out Process • Machine Learning as Black Box Preferences • Productivity and Growth as key IT goals
- 16. TRADITIONAL TECHNOLOGY AND DO-IT- YOURSELF IT • In the 90s and 2000s, we learned skills and thought they would be careers. • Financially, the focus was on maximizing Return on Assets and minimizing Total Cost of Ownership
- 17. REALITY CHECK • Reality Check: no skill lasts forever and dependence on any technology represents fragility and a point of failure. • Do It Yourself only works as long as the employee who built it is still there. But it probably includes • Ridiculous spaghetti code • Lousy documentation • Outdated skills
- 18. IT-BUSINESS ALIGNMENT IS A PARTNERSHIP, NOT A SURRENDER • The answer is often to use any tech that leads to the business outcome • Business must also accept that IT must be repeatable to be truly successful • Outcomes-based efforts lead to “Hero” IT rather than stable IT • Align IT success to long-term growth, not short-term growth or immediate fixes.
- 19. AVOIDING LEGACY IT FRAGILITY • Shift to a Subscription mindset where new tech is a negotiable partnership • Focus on your Return on People, not Return on Assets • Build repeatability and flexibility into IT: Bad things happen when IT relies on Heroes.
- 20. BIG DATA AS AN OPT-OUT PROCESS • In the early days of Hadoop and Big Data, companies focused on ingestion. • The assumption was that data would get sorted out later or that it could just be used in aggregate
- 21. RESULTS OF BIG BAD DATA • Result: Massive amounts of data collected without purpose, leading to massive mistrust of companies and tech • Also, companies often don’t know how to do the graph analytics, machine learning, and data unification needed to get value, anyways
- 22. SOLUTIONS FOR BIG DATA • Use GDPR as guidance for personal data • Focus on collecting data that serves a purpose and can be analyzed by internal personnel. • Build a philosophy for data collection. Start with Why.
- 23. MACHINE LEARNING AS BLACK BOX ISSUES • Unknown algorithms supercede human judgment and stated goals • Correlation and Causation are often confused. • Businesses are complacent about Black Boxes as long as they provide profit
- 24. BLACK BOX INCENTIVES FOR BUSINESSES • In “Business Ethics,” people are taught to do what is right for the business and is simplified to maximizing bottom-line revenue. • But maximizing short-term profit often leads to long-term challenges • Even worse: short-term profit based on unknown data and analytic inputs
- 25. SOLVING THE BLACK BOX • Document data and analytic models used in data science and machine learning. Pure naïve and deep learning is of limited value in today’s IT world • Ask vendors to explain machine learning approaches • Do not take software-provided recommendations for granted • Get educated in algorithms and statistics: our new common language in the digital era.
- 26. PRODUCTIVITY AND GROWTH AS KEY • Productivity and Growth sound positive, but lead to the need for speed • Speed without control leads to mistakes and ethical issues. • Data, analytics, and computing are increasingly differentiated by experience, not performance
- 27. IT FOR THE LONG HAUL • Building for the long haul requires a focus on avoiding fragility, points of failure, and uncontrolled growth and expansion • Solution: focus on solving tech problems with process mapping first rather than simply using brute-force processing and algorithms. • Build room for new technologies, data sources, and APIs to provide additional services within each IT function.
- 28. KEY TOOLS FOR ETHICAL IT • Data Wrangling and Exploration • Data Unification and Catalogs • Expansive BI • Data Science Preparation, Workbenches, and Automation • Guided Machine Learning
- 29. DATA WRANGLING AND EXPLORATION • Your core data must support both structured and unstructured data • Both data and metadata must be available for analysis • A key bridge to managing data across multiple analytic silos
- 30. DATA UNIFICATION AND CATALOGING To contextualize data, IT needs • Shared version of the truth • Consisting of structured and unstructured data • That can be used for BI, analytics, machine learning, and application integration
- 31. EXPANSIVE BI • Business Intelligence solutions best suited for today’s data diversity must • Work with other analytics and BI solutions outside of their own environment • Support a wide variety of data • Be extensible and API-friendly
- 32. DATA SCIENCE AND ANALYTICS • In today’s marketing world, “analytics” solutions are now “Data Science” solutions. But new players have come into the world previously owned by SAS and IBM SPSS. • In general, larger vendors provide a portfolio of tools and services while newer vendors provide focused solutions.
- 33. MACHINE LEARNING • Machine learning is a subset of data science focused on algorithmic models. • These terms can be used interchangeably by laypeople asking for results, which can be very confusing. • Repeatable Machine Learning is best done through Machine Learning platforms rather than statistical workflows
- 34. RECOMMENDATIONS If a trend is threatening your job, learn it before it replaces you. Bring your structured data and your unruly data together so you can at least see it all from one place Break out Black Boxes Don’t define customers and employees as data. Data should be used to help people and achieve specific outcomes
- 35. TAKEAWAY • What will you bring back to your office? • Who will you speak to? • How will your organization benefit from a more ethical IT?
- 36. THANK YOU! For more information on building the business case, evaluating vendors, or working with Amalgam Insights on technology consumption management, please contact Amalgam Insights at: [email protected] Phone: +1 (415) 754 9686 @AmalgamInsights