Online News Association 10 Tech Trends For Journalists September 2016 – Denver
README This is the full presentation deck for Amy Webb’s annual Tech Trends presentation at the 2016 Online News Association conference. This year, Amy’s presentation included the participation of three live bots. You’ll see them in this presentation, but they may no longer be live by the time you try using them.
9 Thank You!
Me: 3 Things 1 Futurist – Technology, Media and Society Founder of the Future Today Institute. We answer “What’s the Future of X” for Fortune 500 + Global 1000 companies, universities, government agencies, large foundations, media companies. Adjunct Prof. @ NYU Stern School. Just completed a research fellowship at Harvard. 2 Former ONA Board Member Served on the ONA Board of Directors for 4 years. Co-organized the ONA conference in 2008. 3 Social Handles #ONA16trends –– @amywebb on Twitter –– @FuturistAmyWebb on FB
YOU: 3 Things 1 There are 10 trends, but I only have time to show 5. You can download a full report at the end. 2 I have some special guests joining me today. You can interact with them during the presentation. 3 Things are going to get weird. Make sure your devices are out >> Twitter, Facebook, SMS... and a Browser window
A media executive pitched me a new strategy for the future of news. This was our exchange:
Exec: “We’re going to be the Uber for news.” Me: (Confused) Exec: “Three years from now, we’ll have an on-demand news platform for Millennials. They tap a button on their phones and they get the news delivered right to them, wherever they are. This is the future of news!” Me: “Is it an app?” Exec: “Maybe. The point is that you get the news right away, when you want it, wherever you are.” Me: “So you mean an app.” Exec: “Yes!” But more like Uber.”
There were TWO crimes committed that day.
Crime #1: The executive mistook a thriving platform for an “Uber for X” tech trend.
What is a real trend? 1. A trend is driven by a basic human need, one that is catalyzed by new technology. 2. A trend is timely, but it persists. 3. A trend evolves as it emerges. 4. A trend can materialize as a series of un-connectable dots which begin out on the fringe and move to the mainstream.
Trends come from these 10 sources of change.
Every journalist pay attention to the future of media and news. Which means following emerging trends. Trends are signposts that can be tracked as they move from the fringe to the mainstream.
Future Today Institute’s Process data > pattern recognition > modeling > interrogating > mapping > scenarios > strategy > pressure-testing
Crime #2: The executive was blinded by Uber’s success and wanted to copy it. So he didn’t notice the real trend worth watching.
Uber’s platform is a long-term threat to traditional media.
What is the one media source we use in cars?
UBER KILLS RADIO NEWS.
Uber + driverless cars will free our attention from captivity. Will you still choose to sit idle, listening to the radio? To podcasts? Or radio apps?
Analogy for the challenge that you all face: Simultaneously thinking about today, tomorrow and many years from now, without getting distracted by shiny objects.
The future of news depends on your ability to see what’s happening at the fringe and follow it to the mainstream.
5Emerging Tech Trends For Newsrooms Scenarios for 2016 and 2026
1 Object Recognition Machines have vision. What do they see?
What do you see?
As humans, we recognize what’s in the picture.
Dry land, tiny shrubs Size of bottles, measured in metric Method of carrying bottles of milk
Probably not in America
Picture = 1 person Abeche = 1 city Chad = one country Africa = 1 continent
7 billion people
Within the next few years, every one of those 7 billion people will own or have access to a camera.
Impending Global Data Challenge Everyday people How do I sort/ store/ find my pictures? How can we use pictures to make sense of the people, Journalists organizations and events that shape our society? How can we use pictures to better understand our Anthropologists evolution? How can we harness pictures to inform us of criminal Law Enforcement wrongdoing? Computer How can we train machines to automatically Scientists recognize objects in all these pictures?
To address that challenge, machines must recognize explicit object... but also context and infer meaning.
Addressing the Challenge: Object Recognition Image credit: Google
Addressing the Challenge: Object Recognition Image credit: Google
Image credit: Google
Google “Josh Hatch Teeth”
Google “Josh Hatch Teeth”
Machines are being trained to recognize objects, and to infer meaning.
Machines can recognize us even when we’re moving...
Today’s face-recognition algorithms... • Recognize you even if you’re moving • Recognize you even if your hair is covering your face Photo credit: Siri Stafford/Getty • ID you by smile, grimace • ID you by posture
Machines are learning to recognize our actions and behaviors, too.
CSAIL: New research - behavior
Machines can recognize our sounds. (Better than we can.)
CSAIL: New research - sound
Here’s how all of this impacts the near-future of newsrooms...
The foundation for computer- generated news videos.
Think of it as a fully-automated version of the tool you’re currently using.
News startups that aggregate and organize media will be in demand.
12 - 36 months Different version for every platform... at full burn, 50-60 videos a day 5 years Different version for every person across platforms... ∞ a day
If you build 1000s of videos, will they come?
Practical Application Start experimenting with object recognition and auto-generated videos. DO NOW: whiteboard your newsroom video strategy for the next 5 years
1 2026: Object Recognition Trillions of cameras, 10x data..... But what about trust?
University of Stuttgart researchers
Smart Dust *computers that are light enough they suspend in the air
What if I hacked the network and changed the metadata: GPS coordinates from Denver to Daytona?
What if our cameras stop reporting the truth?
In 2016: We’re all in the FBI’s Next Generation Identification Interstate Photo System (NGI-IPS)
But, bias in algorithms...
By 2026... AdWords Objects
Filed January 2015 Recognizes objects in video and photos Offers image overlays from brands related to those objects (Like buying Google ads, but visual) Auction system –– companies compete to buy the rights to certain objects, like a coffee cup.
2026: What happens on the 25th anniversary of 911 if marketers buy the object words “Twin Towers” photo credit: Associated Press
Is there an algo that will protect us against our own poor judgement?
2 Crowdlearning Harnessing the data of the crowd.
10 years ago: Crowdsourcing
2006: Crowdsourcing Asking us to contribute content, report on activities, etc. 2016: Crowdlearning Querying our passive data in order to learn or understand something.
Crowdlearning: Possible because of the continual creation and transmission of our data. (Not just phones.)
Simultaneously: Aleppo edits on Wikipedia
Our thinking results in behavior. Our behavior results in data. That data can be used to to learn from the crowd.
Waze also tells us where people are stuck, which implies...
Data from the crowd signals to journalists to ask different questions.
Good Crowdlearning sources Start here: myactivity.google.com Waze Wikipedia HealthData.gov Twitter list data Google’s busy times data iARPA and DARPA’s RFP listings Popular Google Scholar searches and trending papers
Practical Application DO NOW: Make a list of all the sources of data you, personally shed. How can you use those sources for reporting & storytelling?
2 2026: Crowdlearning New stories from old anecdotes. And who owns all that data?
Anonymized crowdlearning data can lead you to new questions about old stories.
Oil and Gas Refineries, too
Neighbors, friends parents, some relatives worked at or near the mills
Cancers that resulted in death My mom - 60 Our next-door neighbor - late 50s My best friend’s dad - 64 Another friend’s dad - 56 Another friend’s dad - 60 Another friend’s mom - 60 My maternal grandfather - 69 My grandparents’ neighbor - early 60s
CDC data: Northwest Indiana doesn’t seem that remarkable
None of the cancers were the same. Not everyone worked in a mill. (My mom was a teacher. Grandparents owned businesses.) But––everyone lived in that area their entire lives.
I know, I know. Correlation ≠ Causation
But it feels weird. Crowdlearning: Something about the cells of people born between 1940 - 1948?
In the next 10 years, we will collect and share health data at an unprecedented rate. What can journalists learn from the crowd’s data?
Reporters will mine this data for numerous stories.
Machine learning and AI will assist in helping to uncover hidden patterns...
...but you’ll need to verify what you’re learning and be transparent about your sources.
Ask questions about all that crowdlearning data:
10 years from now, who actually owns all my data?
Travel Activity (wearable) Health Location: history mobile Search Online Garden Photos & usage sensors Videos Financial Driving Social Posts
What if______ gets sold and auctions off my personal data to a 3rd party? Twitter Spotify Evernote Facebook Netflix Snapchat 23andMe Pandora Eventbrite Google AT&T BlueApron Apple Comcast Amazon Fitbit Box Uber
We should ask and answer that question in 2016, before it becomes a problem in 2026.
3 2016: Mixed Reality (2nd year) Immersive environments changed our expectations for how stories are told.
Mixed Reality (n) Combines the physical and digital realms. Augmented Reality (digital overlays), Virtual Reality (an immersive digital environment) and multidirectional camera angles (360 degree).
MR as 360-degree images + video...
MR without headset...
Hatsune Miku: One of the world’s biggest pop stars
MR with headset...
Mixed reality allows us to connect with news in a more compelling way... ...and is changing our expectations of “storytelling.”
Immersive journalism project from Nonny de la Peña at USC... Features real audio to recreate event outside a homeless shelter...
Use MR to tell the stories that are better experienced than passively watched.
• The refugee crisis • The Great Plastic Reef • Chicago’s gun violence • Global climate change
VR is not a revenue panacea • Not everyone has a headset • “Simulation sickness” • Requires undivided attention • Does your audience have time? • Best place for your resources?
Don’t invest tech and storytelling tools before you plan out your longer- term MR strategy
Practical Application DO NOW: Develop a decision tree for your newsroom to decide when a story is: VR, AR, 360-degree, Hologram, just regular old video.
3 2026: Mixed Reality Because we didn’t plan in advance, we now have serious problems within our digital realms.
First-person, immersive video can also generate empathy in ways we may not like.....
Near-Future: Magic Leap MR combined with our own personal data, artificial intelligence and very fast, cloud-based systems...
See your grandmother sitting on your couch, she sees you on her couch, as if you’re there together Credit system: Pay-per-view news video overlays –– micropayments to see news and information Credit system: Paid personalized news feeds
We didn’t connect the dots in advance. We didn’t ask: • How do virtual worlds change our sense of self? Our sense of belonging to a group? • How do virtual worlds increase/ decrease real-world violence?
• How do we represent religion/ culture/ ethnicity in a virtual world, without using stereotypes? • If while in MR, someone harasses you, or violently attacks you, is it punishable offense? • If someone commits a crime in a virtual world, who’s to blame? Who will help?
4 2016: Conversational Computing Programmable software automates reporting and syndication.
We are entering an era of conversational interfaces. You can be expected to talk to machines for the rest of your life.
Conversations we’ll have...
Near Future Bottable interfaces and platforms will replace standard user interfaces.
Changes the expectations of your news consumers. Conversations replace comments.
These platforms can also simulate a conversation you might have with your editor. To help you think through ideas and concepts.
What’s the big deal? Who’s for and against it, and why?
(Side note) No financial relationships
Akira was created to help you have a conversation to learn more about ONA and #ONA16.
Dmitrii Dumik CEO of Chatfuel
Dmitrii joined Amy on stage to talk about bots and journalism from a developer’s point of view.
Akira can have 4 conversations to help you gain a better understanding of ONA and the ONA conference: 1. ONA 2. ONA conference 3. Bots 4. Futurism
There have been a few early attempts at conversational journalism...
CNN Chatbot: Conversational Search (not AI)
Is this just a gimmick?
Conversational journalism meets people where they already are.
Not a replacement for traditional storytelling. Dynamic Listicle *great for specific subjects like explainers
Basic Corpus: Syria Crisis Explainer Bot (Who, What, Where, When, How, Why?) What’s happening in Syria? Where is Syria, exactly? Can I see a map? When did the crisis start? Why did it start? Who is Assad? Who are the refugees? What caused Syria’s civil war? How many people have lost their lives? What’s the humanitarian crisis, in a nutshell? Are there any peace efforts? How is the U.S. involved?
Challenge to You: Election Night Bot This can be built with Chatfuel (or other) Reminder of who’s running Polling station information (hours, locations) Explainer for local rules, propositions, ballot initiatives What your news org says about the candidates (endorsements) Historical voting trivial/ data Real-Time local candidate results Real-Time state and national candidate results
Amy is the Artificially Intelligent version of me.
Skills: • Specifically programmed to help you understand my ONA16 tech trends • Generative language • Amy learns as they go
Amy’s an experimental bot, created with AI markup language and Pandorabots.
Lauren Knuze of Pandorabots then gave us a detailed backstage tour of the corpus, code and how Amy was built
Challenge to You: Single-Issue Bot This can be built with Pandorabots (or other) • Issue bot: companion to a big story you’ve published • Newsstand bot: allows you have a conversation about the day’s top 3 news stories –– ongoing updates, not a one-off • Local bot: geo-coded; knows basic 311 information; tells you local stories, events, happenings –– ongoing updates, not a one-off
Aikra and Amy are now live, so feel free to interact with them during the session and the rest of the day.
Practical Application DO NOW: Build a conversational journalism project. Try an explainer or single-issue bot. LEARN from your experiment to understand how to meet users where they are to have a conversation.
4 2026: Conversational Computing Mistakes were made.
Basic chatbots require a big database of possible questions and answers.
AI chatbots require data sets, algorithms and training.
Training happens in real-time, and require real interactions with humans.
Other instances of AI behaving badly...
These machines have parents. (programmers) And textbooks. (our data)
We are their mothers and fathers. We are creating them in our likeness.
By 2026, we will realize that we taught machines to talk, and they learned from our structural racism/ sexism/ homophobia/ xenophobia.
Ctrl+Alt+Future Program your bots to help bring awareness to why language matters.
If you try to harass Akira and Amy, they’ve been programmed to teach you how to be a better human.
Botness Scale 1. Perform its designated function well 2. Easy to access and use 3. Help people learn 4. Provide people context 5. Help people learn about their own biases/ broaden their worldviews
REMEMBER This presentation is from the 2016 Online News Association conference, which was held on September 17, 2016. The Twitter, Chatfuel and Pandorabots bots may no longer be live by the time you’re looking in this Dropbox folder.
5 2016: Augmented Journalism (2nd year – was “Cognitive Computing” in 2015) Using new data mining, machine learning and cognitive tools to assist with your reporting.
2001: C.A.R. “Computer-Assisted Reporting”
C.A.R. helps reporters to analyze public documents and crunch data.
Still have to connect the dots to understand the story.
In the 15 years since, machine learning, deep learning, etc. have matured. Systems are more powerful.
C.A.R. v2.0 Augmented Journalism
Augmented Journalism Using computers to help journalists analyze docs and data, connect dots and better understand the subject.
Tokyo, Japan––August 2016 • Female patient in Japan diagnosed with acute myeloid leukemia, a blood cancer
Tokyo, Japan––August 2016 • Female patient in Japan diagnosed with acute myeloid leukemia, a blood cancer • After treatment, recovery was unusually slow
Tokyo, Japan––August 2016 • Female patient in Japan diagnosed with acute myeloid leukemia, a blood cancer • After treatment, recovery was unusually slow • Doctors started wondering if it was actually something else... but in all their medical records and data, they didn’t see anything else
Tokyo, Japan––August 2016 • Female patient in Japan diagnosed with acute myeloid leukemia, a blood cancer • After treatment, recovery was unusually slow • Doctors started wondering if it was actually something else... but in all their medical records and data, they didn’t see anything else • They fed her data into Watson, which crunched her genetic data against a bigger database. 10 minutes later –– new diagnosis: rare secondary leukemia
Tokyo, Japan––August 2016 • Female patient in Japan diagnosed with acute myeloid leukemia, a blood cancer • After treatment, recovery was unusually slow • Doctors started wondering if it was actually something else... but in all their medical records and data, they didn’t see anything else • They fed her data into Watson, which crunched her genetic data against a bigger database. 10 minutes later –– new diagnosis: rare secondary leukemia • Changed her therapy, she’s now in remission
Similar problem for journalists: Connecting the dots based on fact and observation, not just hunches.
Some IBM Watson news experiments...
News Explorer 300k pieces of content 70k sources looks for concepts rather than keywords, in order to make connections between nodes
Visualizing connections between nodes mitigates our own belief bias.
Practical Application DO NOW: Try some of these emerging augmented journalism tools. Learn how to use them to help your newsroom add context and make connections between nodes.
5 2026: Augmented Journalism Soon, you’ll be able to debate your stories with an artificially intelligent editor.
What if you had an artificially intelligent digital editor, to help you evaluate all sides of your reporting?
No, Wikipedia isn’t trustworthy. This will eventually work with proper data sets.
What if debater could be used in real-time against, say...........
IBM Research: Debater
6 Organizational Doxxing What happens when hackers go after the personal details of every journalist and staﬀ member in your organization?
Trend #6: Organizational Doxing In the wake of the Edward Snowden leaks, we’ve seen a number of data dumps. WikiLeaks has published troves of data. Hackers broke into Hacking Team, publishing a massive amount of internal data. Sony has been breached, and so have various branches of the U.S. government. This isn’t about stealing credit card information, but rather about making public the personal details of individuals. “Doxing” is mining and publishing personal information about a person––organizational doxing is when this happens to an entire company. (Security expert Bruce Schneier has written extensively about this.) Why this matters for journalists: In the spring and summer of 2016, hate groups began specifically targeting Jewish journalists on Twitter. In our ever-polarizing political climate, news organizations ought to shore up security and to develop a risk management plan should they find themselves doxed.
7 Digital Frailty We’re creating an unprecedented amount of digital journalism, that’s free and available to everyone...and it won’t be here forever.
Trend #7: Digital Frailty We are creating troves of digital content, only some of which is being archived. Perhaps not every Facebook post should be saved in perpetuity, but might we need to look back on this moment in time and reflect on how our language––how the very way we communicate––was shaped by our Instagrams, our Snaps, and our tweets? Will our future historians look back at this moment, marveling at the amount of anthropological data we were simultaneously creating––and destroying?
Why this matters for journalists: Several news organizations have moved to digital-only publications as a cost-cutting measure. What happens if/when all of the money runs out? We’ve already seen several archives go dark, including a few alt-weekly newspapers, who had produced important investigative reporting series about our cities. That’s today. What will a future society look like if our current media landscape goes dark? Do we have an obligation to preserve our digital journalism? Note: ONA Board Member Mario Tedeschini-Lalli has also written and talked extensively about this, and he’s a good source on this trend.
8 Verification News organizations should be transparent when technology is used extensively for reporting or for story production.
Trend #8: A Nutritional Label for News (Verification) We are increasingly relying on data, algorithms and machine learning for various aspects of news gathering and publishing. News organizations must commit to transparency. They must tell consumers where the information came from, how it was parsed, and how the final story was ultimately produced. Better, still, if that process is divulged in a public place, like Github. Why this matters for journalists: There are too many instances of bias in algorithms to list. Just as consumers expect to see a byline on stories, because it creates a chain of accountability, they will soon expect to know how automated stories and augmented journalists did their work. It is in the public’s best interest, and in the best interest of newsrooms, to create a nutritional label of sorts for their stories, explaining which technologies and datasets were used.
9 Limited Edition News Products Certain digital products are better for stories and consumers.
Trend #9: Limited-Edition News Products (Live SMS) Not every news product needs to last indefinitely. Some organizations have begun to experiment with temporary products: limited-run newsletters, podcasts that only last a set number of episodes, live SMS offerings that happen only during events. Limited-edition news products doesn’t necessarily mean creating a bunch of labor- intensive one-offs. Rather, they can be templates that your organization can use, iterate on and redeploy again and again. Why this matters for journalists: Whether it’s a planned news event (like the 2016 Election), an annual conference (CES, SXSW), a season (basketball, football, winter weather), or a big story that has a defined beginning and end, limited-edition news products are a smart way to deliver content, in a format that works best for the consumer.
10 Journalism as a Service It is time to completely re-imagine how to sustain the future of quality journalism.
Trend #10: Journalism as a Service (JaaS) The central challenge within newspapers is that there are immediate, acute problems––but reasonable solutions will require long-term investment in energy and capital. The tension between the two always results in short-term fixes, like swapping out micro-paywalls for site-wide paywalls. In a sense, this is analogous to making interest-only payments on a loan, without paying down the principal. Failing to pay down the principal means that debt––that problem––sticks around longer. It doesn’t ever go away. So publishers must learn to adopt a new kind of strategy––one which addresses both those immediate financial needs while simultaneously addressing the problem of financial sustainability in the long term. To be fair, this is difficult to do. How does one prove out a positive ROI for something that won’t happen for a few years? Publishers and their investors must get comfortable taking a risk on real R&D––the kind that may not yield any result for a few years. The most important revenue generators for the longer-term has to do with what I call “Journalism as a Service.” How can news orgs become the modern information layer that powers all facets of our everyday lives, whether we’re at home, or at work, or riding around in a driverless car? How can news orgs be used by others in the knowledge economy, such as universities, legal startups and data science companies? How could we rethink news in different kinds of parcels, which include stories, but also include things like verified, searchable databases of people and organizations? Or a service that auto-generates a short report of the opinions on a particular subject, along with a list of quoted experts? Or a calendar plug-in that summarizes the most important news events to pay attention to during the week? All of these services could work outside of the social media landscape, and could be monetized. Why this matters for journalists: Because it is time to completely re-imagine how to sustain the future of quality journalism.
Why isn’t Artificial Intelligence a trend??
AI will impact all aspects of journalism...
Journalism Jobs of the Near-Future OUT IN Data & Algos Investigations Team Reporter Enhanced Reporter Social Media Manager Augmented Reality Producer Web Content Producer Bot Developer Principal Researcher, Media Lab Multimedia Designer Ecosystem Manager Graphic Designer Platforms Manager Public Editor for Code Copy Editor/ Fact Checker Lead Data Scientist Line Editor Automation Experience Designer
AI hasn’t fully arrived. We’re just at the beginning, still. Take time now to learn about what it really is.
AI: What You Should Do Now Learn what AI really is, what it can do, what it can’t do. Learn the lexicon. Read some papers.
The future of new isn’t pre-determined. It’s something we all create together, in the present.
My new book is all about the future, and how you can predict tech trends yourself!
Early access to FTI’s 2017 Trend Report (journalists, teachers and researchers only) https://goo.gl/hvF1ZG
THE FORESIGHT PROGRAM The I Th ns e F tit utur ute’e T s od Te a c y In h T stitut rend e’ s s Annua Forec l R astetain ing er Client Pr Subscript ogr io am n Service
WHAT YOU GET AS A FORESIGHT PROGRAM CLIENT This program ensures that an organization knows all of the near- and mid-future trends on Quarterly On-Site Trends Research Workshops Notes From The Near Future – Monthly Trends Report These half-day customized trends al of the emerging trends you should be aware of, why they matter, and who the key players are driving the trend. Sessions are deeply personalized for your organization and they are highly interactive. First, we detail emerging trends your organization Notes From The Near Future – Monthly Trends Briefing Call needs to know. But from our point of view, information alone isn’t A few days after the report is sent, we host a short virtual hangout That’s why in the second half of to answer any questions you might have. Since al of our clients our quarterly workshops, we apply are invited to join, it’s also a great opportunity to hear from people two of the Future Today Institute’s working outside your industry. tools to help you determine how to move the trend into action. (Even if the result is to simply monitor the trend for a little while longer.) By the Slack Integration end of these quarterly workshops, you’l not only know what trends are emerging, but what to do about For clients using Slack, we include a custom #trends channel to them in the present. continually update and inspire your team. Big Ideas Summit Early Access To Our Tools To Measure Your Invitations Annual Trends Report Forecasting ROI Foresight Program clients receive Our annual tech trends report What’s the value of tracking invitations to our two Big Ideas is covered by dozens of media trends? We’l give you a set of Summits a year. We bring together outlets (Harvard Business tools to help you quantify the the Future Today Institute’s clients Review, USA Today, Wal Street impact of the tracking trends as wel as experts working on Journal, Washington Post within your organization. emerging technologies for an and others) and is typical y exciting roundtable discussion. downloaded more than You’l have the opportunity to meet 250,000 times each year. The and col aborate with bril iant people Future Today Institute clients Ongoing Advising and to share knowledge across receive early access to the industries. (Clients receive 2 seats report, giving them a strategic per Summit.) Summits are typical y edge as they plan for the year The ability to talk with us held in NYC, Boston and DC. ahead. throughout the month for advising, insight and inspiration. (Additional fee applies.) To learn more about becoming a Foresight Program client, futuretodayinstitute.com | 267.342.4300 email us or call 267-342-4300.