Building a defensible AI startup

October 23, 2017 Sam DeBrule

Awesome, not awesome.

#Awesome
“A new study reveals that a machine learning tool can help to identify which breast lesions, already classified as “high-risk,” are likely to become cancerous…The model correctly predicted cancer upgrades in 37 of the 38 lesions or 97 percent. The team found that, had the model been used, it would have helped to prevent nearly one-third of the surgeries conducted on benign lesions.”— Karla Lant, Writer Learn More on Futurism >

#Not Awesome
“Public agencies responsible for areas such as criminal justice, health, and welfare increasingly use scoring systems and software to steer or make decisions on life-changing events like granting bail, sentencing, enforcement, and prioritizing services…but a scoring system used in sentencing and bail by multiple states was biased against black people…. [and] it appears unlikely that the US federal government will join efforts to engage with concerns about the effects and use of algorithms and AI in public life.” — Tom Simonite, Writer Learn More on Wired >

What we’re reading.

1/ With recent advances in machine learning and automation, manufacturing roles transform from robots assisting humans to humans assisting robots. Learn More on The New Yorker >

2/ Conversational AI systems might extend the reach of human caregivers far enough that we could provide elderly people with the care they actually deserve. Learn More on The Washington Post >

3/ In past technological revolutions, low-wage jobs were most threatened by automation — today, algorithms slowly begin to replace traders at Wall Street’s largest firms. Learn More on Bloomberg >

4/ AlphaGo, a program that used machine learning to beat a human world champion at Go earlier this year, is beaten 100 games to none by a new version that didn’t study any human-played games. Learn More on The Atlantic >

5/ In a reminder to take all AI-related news with a mountain of salt, a major Australian publication spins an outlandish tale of our supposedly imminent future. Learn More on Gizmodo >

6/ Polls indicate that Americans enjoy newfound comforts that automation makes possible, but deeply fear the threat it poses to their jobs. Learn More on WIRED >

7/ Machine learning algorithms will eventually become widespread and easy to deploy throughout organizations, making it increasingly important for all employees to understand how and when algorithms should be used to solve business problems. Learn More on Harvard Business Review >

What we’re building.

At work, our inboxes fill up quicker than we can empty them, key decisions are posted and immediately lost in Slack, we forget the thousands of useful articles we’ve read, and we struggle to be present when we’re surrounded by friends and family.

Attention Jacking wreak havoc on our lives and it’s about time we do something about it.

We’re building Journal to make people productive with their information and unglue them from their screens. Join our waitlist, and you’ll be one of the first people to get free access to our product.

Journal is a new way to new way to search and remember information across your work apps and beyond.

Where we’re going.

Highlight from “Routes to Defensibility for your AI Startup”

“…AI brings a new type of network effects that some call “data network effects”. Machine learning algorithms need data to work. While the relationship is not linear, the prediction/classification work done by the machine learning algorithms increases in accuracy as they ingest more data.

As a company adds more customers, it gains more data from each of them to train and refine their algorithms. With more data, the accuracy of the prediction goes up, as well as the overall quality of the product. A better product helps convince new customers to purchase it and contribute their data. The loop is closed.”

Learn More on Machine Learnings >

Louis Coppey
Venture Capitalist, Point Nine Capital

Links from the community.

“Tech giants studying artificial intelligence are enlisting an Obama veteran as their new leader” submitted by Avi Eisenberger. Learn More on recode >

“AlphaGo Zero Shows Machines Can Become Superhuman Without Any Help” submitted by Suneet Bhatt. Learn More on MIT Technology Review >

“The cloud built for machine learning” by Paperspace. Learn More on Paperspace >

“Woebot: AI for mental health” submitted by Samiur Rahman. Learn More on Medium >

“Fundraising, Self-Driving Cars and the Human Condition” by Adam Martell (@adammartel). Learn More on Gravyty >

“A Conversation with AI4ALL on Diversity in AI” submitted by Cielo Lutino (@cielolutino). Learn More on Medium >

“Why 2017 Isn’t Living Up to the AI Hype” by Steven Kuyan. Learn More on Medium >

“Why are we so confident?” by Andrew Mauboussin. Learn More on Machine Learnings >

“Missing Data Filling with Unsupervised Learning” by João Caetano (@jpvcaetano). Learn More on Machine Learnings >

“Facial Emotion Recognition: Single-Rule 1–0 DeepLearning” by Carlos Argueta. Learn More on Machine Learnings >

“Sign Language Recognition with HMM’s” by Ryan Shrott (@ryanshrott). Learn More on Machine Learnings >

“Practical (and simple) implementation of Big Data in Mobile Apps” by Agnieszka Biernacka. Learn More on Machine Learnings >


Building a defensible AI startup was originally published in Machine Learnings on Medium, where people are continuing the conversation by highlighting and responding to this story.

Previous Article
AI and Patient Centric Healthcare
AI and Patient Centric Healthcare

The rapid maturation of AI technologies has driven immense interest in AI, with AI and machine learning (ML...

Next Article
Data is Holding Back AI

I remember grumbling, “Good lord this is a waste of time,” in 1992 while I was working on an AI application...