AI and the Future of Work: Artificial Intelligence in the Workplace, Business, Ethics, HR, and IT for AI Enthusiasts, Leaders and Academics

Otto Soderlund co-founded Speechly in 2016 with Hannes Heikinheimo in their hometown of Helsinki. He believes voice should be a first-class citizen for all apps and making it easy for developers to add voice support from any platform will unlock new innovation.Speechly is a member of the YC Winter 22 batch. Otto and I recently co-presented at the VOICE22 event in Washington DC although I presented remote so this is the first time we’re actually meeting. I heard good things about his talk so I was eager for this discussion. It didn't disappoint.Listen and learn...Why voice is the new app and what it means to develop "voice-first" appsHow RAIN Agency uses Speechly to help auto technicians use voice assistants to fix cars How to accurately detect and transcribe speech when dealing with common challenges like background noise and accentsWhen speech detection achieved "superhuman" levels of accuracyHow Speechly combines speech recognition with natural language understanding (NLU) on the local deviceHow Otto thinks about exercising responsible AIWhy "voice technology won't exist as a separate field in a decade"References in this episode...Responsible AI has a burnout problemAlex Capecelatro from Josh.ai on AI and the Future of WorkKrish Ramineni from Fireflies on AI and the Future of WorkThe Speechly demo site

Jonathan Frankle, incoming Harvard Professor and Chief Scientist at MosaicML, is focused on reducing the cost of training neural nets. He received his PhD at MIT and his BSE and MSE from Princeton.Jonathan has also been instrumental in shaping technology policy related to AI. He worked on a landmark facial recognition report while working as a Staff Technologist at the Center on Privacy and Technology at Georgetown Law.Thanks to great guest Hina Dixit from Samsung NEXT for the introduction to Jonathan!Listen and learn...Why we can't understand deep neural nets like we can understand biology or physics.Jonathan's "lottery hypothesis" that neural nets are 50-90% bigger than they need to be...but it's hard to find which parts aren't necessary.How researchers are finding ways to reduce the cost and complexity of training neural nets.Why we shouldn't expect another AI winter because "it's now a fundamental substrate of research".Which AI problems are a good fit for deep learning... and which ones aren't.What's the role for regulation in enforcing responsible use of AI.How Jonathan and his CTO Hanlin Tang at MosaicML create a culture that fosters responsible use of AI.Why Jonathan says "...We're building a ladder to the moon if we think today's neural nets will lead to AGI."References in this episode...The AI Bill of RightsMosaicMLJonathan's personal site

Eric Olson, CEO and co-founder of Consensus, is a collegiate athlete turned data scientist turned entrepreneur who needed faster access to reliable data while working at DraftKings. Consensus is a search engine that uses a large language model to find answers in peer-reviewed research articles. Eric's living proof that the best entrepreneurs start by solving a problem they've encountered. Hear how Eric's scratching his own itch.Listen and learn...Why Google isn't the answer for scientists seeking evidence-based answers onlineWhy a business model that relies on ads can't solve the "unbiased answer" problem for researchersHow Consensus addresses the problem of conflicting information online from credible resourcesHow to use labels to improve search retrieval accuracy... without introducing bias into resultsHow to use extractive large language models (LLMs), to extract relevant portions of documents and match them to NLP questions Why generative AI like GPT-3 can't answer "what's the consensus opinion out there" when multiple potential answers existWho is responsible if Consensus delivers answers that lead to harmful outcomesWhat Eric learned as a division I NCAA athlete (Go Wildcats!) that has helped him as a high-tech entrepreneurReferences in this episode:Elon Musk launches the Optimus bi-pedal robot at AI dayDan Grunfeld, Stanford athlete and Lightspeed partner, on AI and the Future of WorkConsensus

Mona Akmal, CEO of sales intelligence platform Falkon, is the outspoken co-founder behind an emerging leader in a hot space. Mona migrated to the United States at age 20 with a CS degree and little else. She had an impressive 12-year run as a product leader at Microsoft where she helped scale OneDrive and Office. She subsequently led product and technology organizations at places like Code.org and Amperity. Two decades later, Mona’s the CEO of Falkon AI, an intelligence platform for go to market teams. Falkon recently raised $16M from a group of A-list investors that includes Greylock and Madera among others.Listen and learn...Why Mona's philosophy revolves around two words: "efficiency" and "excellence"What makes a standout sales rep great.How do find signal in noisy sales and marketing dataHow many touches are required from stage one to closing a B2B dealHow to fix the CRM data hygiene problemWhy econometrics approaches perform better than machine learning to solve the "small data problem"Why "everyone needs to be coached and nobody needs to be managed"Mona's (legendary) mental health advice to entrepreneursReferences in this episode...Barr Moses from Monte Carlo on AI and the Future of WorkDerek Steer from Mode on AI and the Future of WorkPeter Fishman from Mozart Data on AI and the Future of WorkStephen Messer from  Collective[i] on AI and the Future of Work Kamal Ahluwalia on AI and the Future of WorkLeading scientists fear AI could lead to nuclear war by the end of the century

Hina Dixit, venture capitalist leading AI investing at Samsung NEXT, grew up in a small town in India from humble beginnings. She couldn’t afford a Starbucks coffee and graduated with significant student debt… which fueled her passion for mentoring and coaching as she became financially independent. Prior to Samsung NEXT, Hina was an Apple engineering leader who helped launch two-factor authentication and other core iOS technologies. Hina’s a reluctant venture investor having always been a builder. A mentor from Homebrew encouraged her to pursue investing and she’s now passionate about finding and funding the next generation of AI and web3 entrepreneurs.Listen and learn… How Hina overcame institutional biases to achieve success in engineering leadership roles and venture investing How being trusted with money at a young age by her father helped Hina become independent and confident in her career The challenges Hina faced transitioning from a builder at Apple to an investor at Samsung NEXT What Hina looks for when investing in AI and web3 startups Where there are opportunities for innovation in web3 and metaverse infrastructure What will prevent Big Tech from centralizing the decentralized web How Hina thinks about responsible AI when evaluating new investments How and when entrepreneurs should engage corporate venture capital (CVC) firms The AR/VR technology Hina wants to invest in… her inbox is open :) References in this episode: Paul Lee, Synesis One CEO, discusses AI, web3 and crypto for gaming on AI and the Future of Work Krishna Gade, Fiddler CEO, discusses AI explainability on AI and the Future of Work Barr Moses, Monte Carlo CEO, discusses data pipeline monitoring on AI and the Future of Work Bindu Reddy, Abacus AI CEO, discusses training and managing data models on AI and the Future of Work How Jack Clark is incorporating AI ethics into new AGI research

Rana Gujral, CEO of Behavioral Signals since 2018, joined the company after a distinguished tech career growing companies like Logitech, TiZE, and Cricut. Behavioral Signals uses emotion and behavioral science to help contact center agents deliver better service. Rana and the team are on a mission to improve customer interactions by using signals other than the spoken word to understand exactly what they need based on indicators like voice tone and pitch.Listen and learn...How to train AI models on past service interactions and outcomes to determine which agents should speak to which customersHow to use deep learning and NLP to process non-speech behavior signals like intonation, pitch, and tonal varianceHow behavior signals can be used to predict stress, duress, and propensity to buy or payHow to achieve high levels of prediction accuracy without processing "the spoken word"Why tone and pitch are better indicators of sentiment than actual words across any language How to compete with Google/Microsoft/Amazon for data when building an AI-first conversational intelligence productThe biggest opportunity Rana sees to use AI to help humans live better livesReferences in this episode:Mahesh Ram from Solvvy (now Zoom) on AI and the Future of WorkGadi Shamia from Replicant on AI and the Future of WorkHow personalization algorithms work in your social feedsBehavioral Signals

Ahmed Elsamadisi built the data infrastructure at WeWork before realizing every company could benefit from his team’s innovation. Traditional star schemas aren’t the best way to manage data. Ahmed instead pioneered a new approach using a single-table column model better suited for real questions people ask. He launched Narrator in 2017 to make it easier to turn data questions into answers and has since raised $6.2M from Initialized Capital, Flybridge Capital Partners, and Y Combinator. Ahmed received his BS in Robotics from Cornell. Hear from a pioneer (and tech provocateur) how new data wrangling techniques are making it easier for mere mortals to get more value out of their data.Listen and learn…How a roboticist who got his start building self-driving cars and designing missile defense systems ended up redefining how data is storedWhy traditional approaches that require SQL to access data are brokenHow a single-column schema eliminates the complexity of joining systems and tablesWhy it’s easier to tell better stories with data using temporal relationships extracted from customer journeysWhy Snowflake, Redshift, and BigQuery are really all the same… and data modeling is the place to innovate What it means to replace traditional tables with activities… and why they’ll eliminate the need for specialized data analysts How to reduce data storage costs by 90% and time to generate data insights from weeks to minutes Why data management vendors are responsible for bad decisions made using your data What is data cleaning and how you should do it What is a racist algorithm Why querying data with natural language will never work Is the WeCrashed version of Adam Neumann’s neuroticism accurate? Hear from someone who lived it... References in this episode:Google’s LaMDA isn’t sentientChandra Khatri from Got It AI on AI and the Future of Work Derek Steer from Mode on AI and the Future of Work Barr Moses from Monte Carlo on AI and the Future of Work Peter Fishman from Mozart Data on AI and the Future of Work Ahmed on Twitter 

Seth Earley is a Chemist by training and an expert on AI. Specifically, how AI is used to improve knowledge management. In fact, he wrote the book on the topic titled “The AI-Powered Enterprise” in which he explains the importance of ontologies when applying AI. Seth is the CEO of Earley Information Science. He has been advising companies on technology strategy since 1994 and is currently focused on AI and knowledge engineering. Listen and learn: Seth’s contribution to AI history… including the term he coined that was co-opted by former IBM CEO Ginni Rometty Why all AI is a data (and information architecture) problem How the Applied Materials field services team reduced time spent finding information by 50% with knowledge engineering and ontologies Why proper information architecture is required for virtual agents to reduce call volume and help live agents What has changed since Seth first published his AI book in 2020 The benefits of semantic search vs. traditional keyword search Where to start with a knowledge management strategy Why “data scientists spend more time being data janitors” How to mitigate the impact of bias in AI training data References in this episode: How AI can detect employee burnoutThe Innovation Delusion on Amazon Earley Information ScienceThe AI-Powered Enterprise on Amazon Kevin Dewalt, Prolego CEO, on AI and the Future of Work

Peter Scott, author, TedX speaker, and futurist, worked at NASA’s JPL laboratory after receiving his Masters Degree in Computer Science from Cambridge. Raising kids made him realize the potential impact of AI to do both good and harm. He left NASA and switched careers to feel confident he was doing all he could to secure their future. He recently published Artificial Intelligence and You after publishing Crisis of Control five years back. Listen and learn: When will we achieve artificial general intelligence (AGI)… and is that the right goal for the AI community? Why we weight the potential of AI doing harm about five times as much as the potential for it doing good. What’s the biggest global problem AI might solve in the near term. How DeepMind’s AlphaFold protein folding technology could change humanity. What does it mean to be human in an era when machines can do more tasks historically reserved for humans? Why Peter blames Big Tech for “breaking” democracy. What Peter expects will be AI’s greatest achievement in the next decade. Why the evolution of a digital race hinges on global economic incentives.References in this episode: Peter’s site: humancusp.comPeter’s podcast: AIandYou.netReducing “radicalization pathways” in social media feedsStuart Russell’s AI textbook  Dr. Mark van Rijmenam on AI and the Future of Work Peter Scott’s homage to Monty Python… and how it relates to AI 

Kia Kokalitcheva, Axios tech reporter, is a Silicon Valley native who writes about tech news and culture. Among other things, she co-authors the popular Pro Rata newsletter (over 200k subscribers) with Dan Primack. Kia has covered many of the most iconic tech stories of the past decade as a writer at Fortune and VentureBeat prior to Axios which was just acquired by Cox Enterprises. Kia recently wrote about Adam Neumann’s new company, Flow. Hear Kia’s perspectives on how Flow could transform living like WeWork transformed working… and why she’s not scared that bots may take her job. Listen and learn… How Adam Neumann of WeWork fame raised $350M at a $1B valuation from A16Z for his new company Flow… before launching  Kia’s proudest moment as a journalist What the acquisition of Axios by Cox Enterprises means for journalism How Flow may be more than the reincarnation of WeWork’s failed WeLive experiment As a culture, are we ready for communal living? What is the future of company perks… are the days of on-site dry cleaning numbered? How the generational shift is impacting cultural norms in the workplace What tasks bots will never do better than live journalists References in this episode: How AI is transforming journalism according to The Knight Foundation Gary A. Bolles discusses the WorkNet on AI and the Future of Work Mark van Rijmenam, The Digital Speaker, on AI and the Future of Work