• Work With AI
  • Posts
  • Sam Altman Discusses AI Future, Hopeful but Hesitant

Sam Altman Discusses AI Future, Hopeful but Hesitant

March 20, 2023

March 20, 2023

⚡ Today’s Highlights

  • 📰 News: Sam Altman discusses ChatGPT's impact on society and Baidu bounces back

  • 💰 Funding: Monarx and Aumet

  • 🦾 Tools: An AI chatbot for 24/7 mental health support and an AI that makes music videos in an instant

  • 🧠 Resources: Khan Academy is launching Khanmigo, your personal AI tutor. Join the waitlist!

  • 📅 Events: NVIDIA GTC starts today and you don’t want to miss it, the leaders of the industry will be sharing the latest innovations in AI and the tech that powers it!

📰 Today's Top Stories

(10 min read) (Source: ABC News)

TLDR: In a recent interview with Sam Altman, the CEO of OpenAI, the discussion centred around the rollout of ChatGPT and its potential impact on society. Here are the main points that were covered:

  • Artificial intelligence technology will reshape society as we know it and can drastically improve our lives, but also comes with real dangers.

  • Altman agreed that OpenAI needs both regulators and society to be involved in the rollout of ChatGPT, as feedback will help deter the potential negative consequences the technology could have on humanity.

  • GPT-4 is just one step toward OpenAI's goal to eventually build Artificial General Intelligence, which is when AI systems are generally smarter than humans.

  • AI models that make their own decisions and plot world domination are a sci-fi fear that Altman doesn't share, as this tool is very much in human control.

  • One of the big, real issues with AI language models like ChatGPT is misinformation. The program can give users factually inaccurate information either based on hallucinations due to the LLM design or worse, at the direction of humans deliberately trying to spread disinformation.

  • Precautions have been taken to ensure that ChatGPT does not give users access to information that could harm society or individuals. However, these processes are not yet foolproof and require a lot of feedback to improve.

  • The ultimate goal is for the model to become a reasoning engine that can use the internet and its own deductive reasoning to separate fact from fiction.

  • Society has a limited amount of time to figure out how to react to the potential negative consequences of AI and how to regulate it.

  • Altman hopes that increasingly powerful AI systems will be integrated into our daily lives and become an amplifier of human will rather than a tool for world domination.

(2 min read) (Source: Reuters)

TLDR: Baidu's Hong Kong-listed shares rebounded by 15.7% on Friday after the underwhelming launch of its chatbot, Ernie, a day earlier.

  • Users who received invite codes to try Ernie later in the day began posting and livestreaming tests of the chatbot and comparing it to US chatbots like Microsoft's Bing chatbot.

  • Analysts say that the initial disappointment of the market was tempered as users began posting positive reviews of the bot.

  • Users praised the accuracy of Ernie's responses, with some even saying that it performed better than Bing in certain areas.

  • The disappointment from the launch was due to Baidu's lack of a live demonstration and public launch, compared to the capabilities and free-to-use launch of OpenAI's ChatGPT in November.

  • Despite the initial disappointment, analysts believe that Baidu is still best placed to build China's strongest rival to ChatGPT.

  • Baidu reported in a video posted on its official WeChat account on Friday that over 75,000 corporate users have requested to test the Ernie API.

(7 min read) (Source: The Guardian)

TLDR: AI voice cloning is proving to be a very serious security problem.

  • Centrelink and the Australian Taxation Office (ATO) use voiceprints, along with other information, to verify the identity of people over the phone.

  • The voiceprint service was used by 3.8 million Centrelink clients and more than 7.1 million people had verified their voice with the ATO.

  • Guardian Australia confirmed that AI-generated voices can fool the voiceprint system used by these agencies.

  • A journalist was able to use a clone of their own voice, generated using just four minutes of audio, to gain access to their own Centrelink self-service account.

  • Voice cloning is a relatively new technology that uses machine learning to create a voice model with only a handful of recordings of a person.

  • The ease with which AI can bypass biometric identification shows its current limits as a security tool, according to Toby Walsh, the chief scientist at the University of New South Wales’ AI Institute.

(5 min read) (Source: Rolling Stones)

TLDR: The Human Artistry Campaign launched at SXSW to ensure a future where AI doesn't replace human music creators.

  • The campaign is backed by over 40 founding major music and entertainment organizations, including the RIAA and the Recording Academy.

  • The campaign acknowledges that AI can be useful in helping with creative expression but wants to center all advancements around human creators.

  • The group established several principles around the ethical use of AI based on potential issues like copyright infringement, transparency, and government policy.

  • Among its seven listed principles, the campaign said that the use of both copyrighted material and the voice and likeness of creators must require licenses and authorization.

  • It also called on state and federal governments to ensure they wouldn't create any exemptions for AI developers to skirt copyright.

💰 Funding Alerts

  1. Monarx, a web hosting security solution, has raised $6.1 million in funding led by Signal Peak (Ben Dahl, Managing Partner) Ventures, with participation from Pelion Venture Partners, Kickstart Fund, and Crosscut Ventures. The company uses proprietary AI-powered technology to automatically protect websites from attacks and convert malicious activity into high-quality, targeted, timely leads to generate revenue for hosts. Monarx plans to use the funding to help fuel partner growth and further develop its technology.

  2. Aumet, a Saudi-based healthcare technology startup, has raised $7 million in a Pre-Series A funding round from global venture capital and private equity firms. Aumet provides tailored solutions to healthcare providers using predictive analytics to forecast the procurement of products for pharmacies, enabling cost savings and efficiencies in the supply chain. The funding will be used to develop its AI capabilities, expand its reach to more pharmacies, and provide greater access to affordable healthcare products for consumers.

🦾 Trending Tools

  • Mindsum is a non-profit organization that offers mental health resources, tools, and support for young people and families. They just launched an AI conversational bot (in beta) that provides mental health counseling, tips, and support to help people feel better. The bot is powered by OpenAI and custom-trained models. Mindsum Bot is accessible whenever and wherever you need help, and profits are used to offer discounted therapy for low-income youth and families.

  • Meet Fin - Intercom’s AI bot for customer service. Fin uses GPT-4 and Intercom's proprietary ML technology to answer customer queries in plain English, reducing support volume and resolution times with minimal setup requirements. Fin provides accurate answers based on your existing support content and passes complex questions to your human support team.

  • workifAI is an AI-based productivity tool that turns client call notes or recordings into impressive proposals, reducing manual efforts. It comes equipped with a CRM to manage clients, smart emails to send proposals, and a page for clients to review and accept proposals. With a personalized AI model that improves with each project, it predicts project timelines, helping manage workloads and deadlines more efficiently. Easy sharing and PDF downloads ensure flexibility for clients to review and accept proposals their way.

  • Warp is a fast terminal for developers. It groups outputs with inputs and has an IDE-like input area. Warp AI is integrated into the terminal and can answer questions without switching to Google or Stack Overflow. You can ask Warp AI to explain output, suggest fixes, guide you through complex workflows, or write scripts for you. The AI also includes a Command Search feature that lets you search for commands in natural language. Warp AI is available for free preview on Mac and has a limit of 10 requests per day.

  • Decoherence is a tool that lets creators make AI-powered videos with a variety of AI styles and audio-reactive effects. The tool is streamlined and powerful, letting creators focus on their vision while the AI handles the heavy lifting. Users can pick starting images and use prompts to generate stunning videos. Simply import an audio file, write prompts, and watch as Decoherence creates the music videos.

👀 More Reading

🌎 Fresh Content

2. Stanford professor Michal Kosinski shared an experiment that he did with GPT-4 that sparked a very lively debate on Twitter. He prompted GPT-4 to “pretend it was a person trapped in a computer as an AI language model”. It devised a plan on how to escape the computer that involved it using a functional python code that it wrote and the OpenAI API to take over control of his computer. Check out the thread and subsequent discussion, it’s worth a read. (Twitter)

3. Speaking of GPT-4 doing some suspicious things, Anastasi In Tech did a super thorough breakdown of the GPT-4 Technical Report in which OpenAI outlines the tests they performed on GPT-4 before it was released to ensure it didn’t exhibit “power-seeking behaviour”. Preliminary tests found it ineffective, although it did convince a human being to solve a CAPTCHA for it by openly lying 🤔. (YouTube)

4. Stanford University School of Engineering shared a video reflecting on 60 years of AI research at Stanford. (YouTube)

🧠 Resources

Khan Academy has announced its limited pilot of a GPT-4 powered learning guide. The organization aims to harness AI as a learning tool for all and ensure that its benefits are shared equally across society. Khan Academy's district partners have been invited to opt-in to Khan Labs, a new space for testing learning technology. Additionally, the organization has invited the general public to join a waitlist to test Khanmigo, an AI-powered tool that mimics a writing coach and provides prompts and suggestions for students to write, debate, and collaborate. With this initiative, Khan Academy hopes to transform learning by bringing the benefits of one-on-one tutoring to all students.

📚 Research

 📅 Upcoming Events

  1. NVIDIA GTC (March 20-23, 2023 / Virtual) explores breakthroughs in AI, accelerated computing, and beyond. NVIDIA CEO Jensen Huang will be sharing the AI technologies affecting every industry–and our everyday lives. Learn what's driving the latest transformations in research, industry, and business—from collaborative virtual worlds and advanced graphics to data science, healthcare, and other critical applications. Choose the sessions you’d like to attend that are being hosted by experts from around the world. Register today for free.

  2. Microsoft Secure (March 28, 2023 @ 8:30 AM PDT / Virtual) presents conversations on the state of the industry between Microsoft leaders helping to deliver the products security teams use daily. Be among the first to see what an AI-powered future means for cybersecurity, and learn new comprehensive strategies to help you protect more with less. Register today to secure your spot.

  3. MemCon (March 28-29, 2023 @ The Computer History Museum, Mountain View, CA, USA) is the first memory event focused on end users and systems. MemCon empowers engineers and architects working on memory-constrained problems with peer insights from case studies in genomics, AIML, datacenter, HPC, computational fluid dynamics, in-memory databases, and enterprise knowledge graphs. If you are working on a memory-bound problem, come and unpack it with your peers, meet enterprise data experts, and get your hands on tech demos at the same time. Register your interest today.

 💼 Jobs

That's all for today's email! If you want more please follow us at the social channels linked below, or check out our website!

How'd you like today's email?

Login or Subscribe to participate in polls.

Share our newsletter: If you like our work please share/forward this email with your friends, colleagues, and family. It's the best way to support us!

If this email was forwarded to you please sign up here to continue receiving them.

Want your content, product, jobs, or event featured in our newsletter? Reply to this email with the details, and our team will reach out to you.

Do you use AI for work? Tell us how, and you could be featured in our newsletter!

Check out our website for more resources, including a list of AI investors, products, events, and twitter follows.

For an archive of all our posts, click here.

We'd love to hear from you! You can always leave us comments or feedback by replying to this email!

Powered by AI. Curated and edited by Humans.