Sign In

Olya Gurevich

Co-founder & Chief Scientist at Marvelous AI

Olya Gurevich is a co-founder and Chief Scientist at Marvelous.ai, a company focused on artificial intelligence.2 She has a diverse background that combines linguistics and technology, with a particular emphasis on applying her skills for social good.1

Professional Experience

Before her role at Marvelous.ai, Olya held the position of Engineering Manager at Apple, specifically working on Siri.2 This experience in a major tech company likely contributed to her expertise in AI and natural language processing.

Education and Career Transition

Olya's educational background is in linguistics, but she has since transitioned into the technology sector.1 This unique combination of linguistics and tech expertise has shaped her career path and likely influences her approach to AI development at Marvelous.ai.

Current Focus

At Spotify, Olya is involved in ensuring the safety of content on the platform.1 This role aligns with her interest in using technology for social good and demonstrates her commitment to addressing important issues in the digital space.

Location

As of 2019, Olya was based in San Francisco, California.2 This location is consistent with her involvement in the tech industry, as San Francisco is a major hub for technology companies and startups.

Olya Gurevich's career trajectory showcases a blend of linguistic expertise, technological innovation, and a commitment to using AI for positive social impact. Her role as Co-founder and Chief Scientist at Marvelous.ai likely involves leveraging this unique skill set to drive the company's AI initiatives and research.

Highlights

Apr 10 · linkedin

Emerging AI capabilities and their consequences (good or bad) are a hot topic of discussion. But it is just as important to talk about #incentives, or why individuals or organizations might choose to use AI in certain ways. AI can be transformative for human experience and quality of life, but only if incentives (both short-term and long-term) for its use are aligned with benefits to humanity.

Some examples of negative incentives and how to address them:   Actor: State actors and ideological groups Incentive: Geopolitical influence for or against something. This includes the use of extra-persuasive content or deepfakes to sow malicious narratives, undermine support and trust in democratic institutions and weaken social cohesion: https://lnkd.in/gCbvdkb7 Modification: Make it prohibitively expensive or inconvenient to use AI for these purposes, through legal consequences, content moderation, or counterspeech. The implementation burden falls on governments, content platforms, and community organizations.

Actor: Criminal or unscrupulous organizations Incentive: Financial gains from scams, ad-monetized websits, or product sales. This includes phishing content, “Nigerian prince” letters or AI-generated SEO: https://lnkd.in/gFJDjhtt Modification: These incentives are always going to exist; protection against them needs better cybersecurity tools, implemented and deployed by consumer-facing web surfaces.

Actor: Commercial enterprises Incentive: Competitive advantage and increased shareholder value. At worst, this can lead to deceptive or discriminatory business practices, or hasty deployment of untested AI to customers, premature restructuring of teams: https://lnkd.in/geS2Tb3c. Social media platforms might tolerate or amplify harmful AI content because of its high engagement. Modification: The drive for advantage can also be good, when aligned with public opinion or customer sentiment. Trustworthy and safe AI can make products more usable, attracting more customers. Government-led compliance requirements can also create positive incentives, like for food or car safety. And sometimes, a punitive legal strategy (e.g. tobacco or opiate lawsuits), incentivizes companies to behave better.

Actor: Individuals Incentives: Faster completion of work tasks, improved social status, revenge against perceived slights, or exploitation of the vulnerable. At worst, these can lead to cheating, misrepresentation of one’s identity, slander, deep-fake pornography, or AI-enhanced grooming. Modifications: Some of these behaviors are illegal and should be prosecuted. But the urge to improve one’s work performance or social status can be a good thing. If AI tools are designed to enhance human productivity while rewarding our creative impulses, and feel fun, joyful, and satisfying to use, people will be more likely to employ them to good ends.

Would love to hear more about improved incentives for better AI use!

Mar 5 · linkedin

Generative AI has the power to transform the world, in ways both good and bad.  A lot has been written about its potential risk for societies as well as individual people (as consumers, students, medical patients, etc.).  But rushing untested, non-trustworthy AI into production is also bad for business; it can lead to embarrassment, reputational harm, and negative financial outcomes.  Here are just a few recent examples of #genai #trustandsafety screw-ups.

[update: an AI generated ad gone terribly wrong: https://lnkd.in/gMCzVhbU]

  • A court forced Air Canada to respect a bereavement policy invented by customer service chatbot: https://lnkd.in/gzHCGTjd

  • Hacker convinced a GM chatbot to sell him a car for $1: https://lnkd.in/ghwwbsnu

  • Denial of Wallet attacks: a user can run up a huge OpenAI bill for a company when prompts are not secured: https://lnkd.in/gcXPUY8b

  • TurboTax, H&R Block deploy AI-assisted tax advice, leading to embarrassingly wrong results and potential audits for customers who follow the advice: https://lnkd.in/gYq-En-4

  • Sports Illustrated forced to admit using fake writers, AI-generated stories of poor quality, executives fired in the wake of the scandal: https://lnkd.in/gQf3PFvu

  • The National Association for Eating Disorders took down a chatbot that gave harmful advice to users: https://lnkd.in/gZ3zZFce

What are your favorite examples?  And if AI is part of your business strategy and you’re worried about potential risks, please reach out!  It’s much easier to incorporate #responsibleAI at the outset than to bolt it on after a product is well underway.

Related Questions

What is Marvelous.ai's main focus and mission?
How did Olya Gurevich transition from linguistics to technology?
What are some notable projects Olya Gurevich has worked on at Spotify?
What is Olya Gurevich's educational background?
How did Olya Gurevich become an Engineering Manager before co-founding Marvelous.ai?
Olya Gurevich
Olya Gurevich, photo 1
Olya Gurevich, photo 2
Add to my network

Location

San Francisco Bay Area