How and why non-profits can deploy artificial€¦ · Predictive Analytics – “What is most...
Transcript of How and why non-profits can deploy artificial€¦ · Predictive Analytics – “What is most...
How and why non-profits can and should deploy artificial intelligence today
What is Artificial Intelligence?
The capability of a machine to
imitate intelligent human
behaviour
What is Machine Learning?
– Machine learning is an application of artificial intelligence
that enables systems to automatically learn and improve
from experience without being explicitly programmed
– Machine learning focuses on the development of computer
programs that can access data and use it to learn for
themselves
So…
– We talk a lot about AI
– We really need to deploy machine learning
– Which is semantics, the difference doesn’t really matter to us
now
– We need to simplify the narrative so we can access the
benefits
– We need to talk about Data (Quality & Governance)
What can AI do for me?
– Discover insights by comparing some
form of input with data they already
have
– Do statistical evaluations to predict
outcomes
– Recommend next steps based on the
results of their evaluations
– And some AI applications will
automatically implement actions
themselves
Why does this help?
– Do you have a mission-related question
to ask and have answered by your data?
– Do you have access to usable data?
– AI/ML gives you the ability to use your
data at scale, at pace
Should I?
– Don’t just think you have to
– Don’t do it because others are
– Don’t do it then work out why
– Learn from others
– Agree objectives and measures
– Trial, learn, commit
Routine
administration
– Chatbots!
– customer service and routine request
– manage first-line support queries and direct
queries to (the right) humans as needed
– automate repetitive tasks, reducing the risk of
input errors, accelerating data collection and
delivering a consistent experience
Saving you time
– you and your team can focus on what you do best
– AI handles what it does best, the time-consuming
work of data entry and analysis.
– You spend less time dealing with the tedious
details of scouring data and manually
implementing technology-embedded processes and
more time doing meaningful work
Supporting the cause
– Access to more sophisticated metrics
– Often displayed better
– To better understand audience and impact on
attitudes and behaviours
– Metrics aren’t everything, but they can help to give
an overview of the bigger picture at times
– New ways of displaying information may help
constituents, members or funders to better connect
Finance & Governance
– Fraud and corruption are major challenges
– Impossible to monitor every financial transaction and business contract.
– AI tools can help managers automatically detect actions that warrant additional investigation.
– AI and ML create early warning systems, spot abnormalities, and thereby minimize financial misconduct.
– These tools offer ways to combat fraud and detect unusual transactions.
Animal Welfare
– PAWS, an organisation dedicated to combating
poaching
– Using modelling and machine learning to give
park rangers the information they need to predict
poachers’ actions and stop them.
Fighting Abusive
Behaviour
– Online trolls disrupt organisations and target
particular individuals.
– Amnesty International pioneered a machine
learning and crowdsourcing tool that can spot
“online abuse automatically” and enable
organisations to remove it.
– The Troll Patrol can identify racist, sexist, or
homophobic tweets, among other objectionable
content and eliminate the abuse.
Saving vulnerable lives
– Crisis Text Line still implements a human-to-human
volunteer model, but has the largest open source
database of youth crisis behaviour in the US
– It uses AI to dramatically shorten response time for
high-risk texters from 120 seconds to 39.
– Crisis Text Line leveraged ML to identify the term
“ibuprofen” as 16 times more likely to predict the
need for emergency aid than the word “suicide.”
– Now using AI, messages containing the word
“ibuprofen” are prioritized in the queue
Predictive Analytics
– “What is most likely to happen based on my
current data, and what can I do to change that
outcome?”
– Use historical data, machine learning, and AI to
predict what will happen in the future.
– Minority Report was released in 2002…
Trend Analysis
– Passive
– Not answering questions, but finding trends
– As data is added, built up, trends are identified
– Not constrained by parameters we set or
subconscious predeterminations
Personalisation
– Looks at past interactions, e.g. what they viewed on your website, links clicked in emails and social updates etc.
– Analyse data, spot patterns, and learn what works best for different segments of your audience
– Craft your appeals and communications based on what prospects or donors care about most.
– Email opens will increase and unsubscribes will decrease when you know
– What’s the best time to ask a one-time donor to give again?
– What’s the best communication channel to use with a particular donor?
– Which stories resonate most with a particular donor?
– Passive evidence-based, not assumed / expected, and continually tuned
Engagement Scoring
– Currently we set metrics, assign scores, measure
– AI can use more evidence to do this better
– To better assign value of activity in line with
outcomes it can see
Fundraising
– The key to successful non-profit fundraising and AI is teamwork.
– Now human fundraisers team up with AI assistants, with each half doing what they do best.
– The AI assistant develops and applies algorithms to ingest, clean, enrich and search vast amounts of data to recognize patterns and give non-intuitive recommendations.
– The human fundraiser applies judgment and context to select from those recommendations to reach out to the right donor with the right message.
Knowledge Management
– The ability to consume vast quantities of data in an
instant
– To profile match the requester with deep mining of the
information
– Long been the key to membership recruitment and
retention
– The Holy Grail for esteemed membership bodies who are
now having to compete with Google to be the font of all
knowledge
– Data Governance and Quality becomes the imperative
Preventative Diagnostics
– AI (IBM’s Watson) saved a woman’s life through early
diagnosis of a rare form of leukemia by comparing her
genetics to 20 million oncology studies. In 10 minutes!
– King’s College is revolutionising the radiotherapy industry
and treatment process. AI will be used to bolster an existing
workforce of trained medical professionals.
– An AI is now capable of detecting the early signs of
Alzheimer's Disease; picking up early signs an average of 6
years before human physicians are able to issue a diagnosis.
– An AI learnt how to model and predict heart disease
mortality rate in patients.
“Deep Patient”
– Analysed 700,000 patient records, with no framework
understanding to hang it all on
– Was able to make more accurate diagnosis than human
physicians.
– Algorithms work because they capture, better than any
human can, a universe in which everything affects
everything else, all at once.
Digital Health Assistant
– Reason Digital, Parkinson’s UK, the Stroke Association,
Muscular Dystrophy UK and the MS Society
– Transforming the way medical advice and information is
delivered to almost half-a-million people in the UK.
– Use machine learning to
– develop an understanding of the person being supported
– adapt to their needs over time based on interactions.
– DHA will provide emailed content and support specific to
the individual’s needs, making it more effective
What does this tell us
– There are lots of different applications of
the same core capabilities and principles
which can help you
– Achieve your mission
– Serve your members or supporters
– Support your beneficiaries
– Do some good
Yes you can!
– You need mission-related questions to
ask and have answered by your data
– You need access to usable data
– You need expertise to build the
algorithm for predictive models
– You need a strategy and plan to manage
ethical concerns around privacy & data
bias
Next steps…
– Start tracking and collecting data
– Ensuring that the data going into your
systems is clean and deduped
– Develop a strategy for the what and why
– Agree your priorities
– Initiate a change programme