Contrary Research Rundown #30
The AI safety debate, and new memos on Mercury, Wealthfront and more
Research Rundown
The AI safety debate has recently sparked meaningful discussions about the risks and opportunities associated with advanced AI systems. As these technologies continue to develop, we must consider the potential consequences and take steps to mitigate any potential harm. OpenAI’s Sam Altman and the AI researcher Eliezer Yudkowsky discussed it on Lex Fridman’s podcast a few weeks ago. Nat Friedman, ex-CEO of GitHub, talked about it on The Lunar Society last month. Tyler Cowen argued the benefits outweigh the risks in Marginal Revolution. Investor Elad Gil wrote an essay about the potential threats.
One of the key concerns raised in the AI safety debate by AI pessimists and alarmists is the potential for these systems to act in unintended ways. With the ability to learn and evolve on their own, AI systems may make decisions that are unpredictable or outside of the intended scope of their programming. Another concern is the potential for AI systems to perpetuate existing biases or discrimination. If these technologies are trained on biased data or designed with implicit biases, they may make decisions that reflect and perpetuate these biases, leading to harm or unfair treatment. AI safety is about identifying and mitigating potential risks, improving the safety of AI systems, and preventing unintended harmful consequences of AI. There have recently been calls for more transparency in AI, which may improve trust and accountability.
On the other hand, AI optimists argue that the debate may slow the development and deployment of AI systems, potentially delaying the benefits these technologies could provide. Some argue that the risks associated with advanced AI systems could be clearer and overstated and that the focus on safety could lead to unnecessary regulations or restrictions. While this debate is going on, startups are turning to AI. Y Combinator’s latest class reflects that: 91 startups, or 34%, of the current YC class, say they are an AI company or use AI in some way. Examining YC batches from previous years, we can observe how venture capital trends have changed. The current cohort has fewer Web3 and crypto startups than in 2021 and early 2022. Many believe that AI will be a transformative technology unlike anything we’ve seen before.
Doctolib is an all-in-one platform for healthcare professionals, covering appointment scheduling, telemedicine, messaging, and back-office operations management. To learn more, read our full memo here.
AirGarage is a modern parking operator that helps parking lot owners monetize their asset by handling enforcement, operations, software, payments, and more. To learn more, read our full memo here.
OpenStore provides liquidity for ecommerce business owners by automating the pricing process and giving the owners an offer within one day for a payout. To learn more, read our full memo here.
Wealthfront is a robo-advisor that offers a combination of financial planning, investment management, and banking services through its platform. To learn more, read our full memo here.
Mercury provides a modern banking experience to startups and offers FDIC-insured accounts, debit cards, currency exchange, and wires. To learn more, read our full memo here.
Quora’s Poe will let users make their own chatbot using prompts combined with an existing bot, like ChatGPT, as the base.
Microsoft Teams users will get access to a collection of 26 popular Lenses from Snap during meetings.
The founder of Tonal Systems stepped down as CEO after the at-home fitness company completed its latest funding round at a fraction of its prior valuation.
Betterment, Wealthfront, and others offer higher returns and more protection than most bank accounts. Check out our memo on Wealthfront.
a16z released its 2023 State of Crypto report. Stay tuned next week for our latest deep dive on the history of crypto and decentralization.
The Commerce Department begins exploring possible rules for ChatGPT and other generative AI tools, requesting comment from the public.
Visa partnered with PayPal, Venmo, and others to power interoperable digital payments.
TikTok sister app Lemon8 surges in the US.
Substack rolled out Notes, its Twitter-like service, to all users. Check out our memo on Substack.
OpenAI partnered with Bugcrowd to launch a bug bounty program, offering rewards from $200 to $20K. Check out our deep dive on OpenAI.
New York Times story on the impact of recent AI developments on the legal profession.
Strava announced a Spotify integration so users can access their Spotify favorites within the Strava app during workouts. Check out our memo on Strava.
WhatsApp now lets users shop and pay in-app in Brazil.
Universal Music Group wants the music streaming platforms to block AI services from scraping melodies and lyrics from their copyrighted songs.
Silicon Valley VCs tour the Middle East in the hunt for fundraising.
An interview with the ex-President of global sales at Tesla and former COO of Lyft on electric cars and autonomous driving.
Databricks released Dolly 2.0, the next version of its LLM, and a dataset trained on 15K records generated by its employees. Check out our memo on Databricks.
Bessemer Ventures released the State of the Cloud 2023 report.
Discord is cooperating with US law enforcement’s investigation into the classified material breach. Check out our memo on Discord.
AWS began offering customers access to LLMs by Anthropic, Stability AI, and AI21 Labs.