Image created by MullenLowe for MassArt campaign Defining an Artificial Intelligence strategy SSP London, 5 th July 2018 Isabel Thompson Senior Strategy Analyst, Holtzbrinck Publishing Group
Scholarly Communication doesn t have an AI problem It has a strategy problem
Problems Competing investment priorities Unclear use cases Unknown ROI Not enough data Poor quality or siloed data Un-agile organizational set-ups A lack of data scientists and tech talent
Outline 1. What is AI? 2. What can it do? 3. What should I do?
1. What is AI?
AI is The field of computer science dedicated to solving cognitive problems commonly associated with human intelligence, such as learning, problem solving, and pattern recognition - Source: https://aws.amazon.com/machine-learning/what-is-ai/
AI is an intuition machine
Submarines don t swim https://www.youtube.com/watch?v=k1bfwrgvqmc
2. What can it do?
Human-in-the-loop No human-in-the-loop Adaptive Hard-wired Assisted Intelligence Help perform tasks quicker or better Augmented Intelligence Augment human intelligence, and learns from interactions Automation Manual, repetitive tasks Automation Intelligence Can adapt and act autonomously Adapted from source: https://www.pwc.co.uk/economic-services/assets/ai-uk-report-v2.pdf
Natural Language Audio & Speech Machine Vision Navigation Visualization Sensory Layer Robotic Process Automation Deep Question & Answering Machine Translation Collaborative Systems Adaptive Systems Behavioural Layer Knowledge Representation Planning & Scheduling Reasoning Machine Learning Deep Learning Cognitive Layer Statistics Econometrics Optimization Complexity Theory Computer Science Game Theory Foundational Layer Adapted from source: https://irpaai.com/ai2017-recap-ny/resources/anand_rao.pdf
Non-industry example Non-industry example Scholarly Comms example Classification Anomaly detection Clustering Continuous estimation
Non-industry example Non-industry example Scholarly Comms example Recommendations Ranking Other optimization Data generation
3. What should I do?
Don t invest in AI! Invest in a business problem
Use in the core parts of value chain Predict future trends, sales, R&D Produce products more efficiently Promote right products, people, & time Provide an enriched, personalized UX Editorial & Production Sales, Marketing, Discoverability https://www.mckinsey.com/~/media/mckinsey/industries/advanced%20electronics/our%20insights/how%20artificial%20intelligence%20can%20deliver%20real%20value%20to%20companies/mgi-artificial-intelligence-discussion-paper.ashx
Top line effects Changing sources of value Bottom line efficiencies
Competitive advantage Data
Partnerships Will need to be more creative and open Because most AI talent sits with vendors Requiring different relationship structures
Buy or build? How essential the data is for competitive differentiation Build Buy How unique your data set is vs vendors https://www.youtube.com/watch?v=k1bfwrgvqmc
McKinsey s 5 aspects of a successful AI transformation 1. Have a business case 2. Build a data ecosystem 3. Get the right tools and techniques 4. Integrate throughout your workflows 5. Adopt an open culture and reskill the workforce https://www.mckinsey.com/~/media/mckinsey/industries/advanced%20electronics/our%20insights/how%20artificial%20intelligence%20can%20deliver%20real%20value%20to%20companies/mgi-artificial-intelligence-discussion-paper.ashx
What if we think of AI as an AI-lien pentopus?
Get the right business case
Build a data ecosystem
Make sure you have the right tools, people, and techniques
Integrate throughout your workflows
Adopt an open culture and reskill the workforce
Summary 1. AI can do a lot of stuff 2. But you need a business case, and AI is not always the answer 3. We need to think differently about competitive advantage, partnerships & vendors 4. And change our organizations structure, people, processes, data 5. Because the way businesses and research work is going to look very different
Thank you! Isabel Thompson isabel.thompson@holtzbrinck.com