AI is a rapidly developing sector that Insight has invested more than $4B since 2014. We’re currently tracking 23,315 AI/ML startups and ScaleUps. Naturally, our team has seen a lot, and the discussions around this topic generate a lot of opinions.
Here’s a look behind the curtain into some of what we’re reading, sharing, and discussing across the team at Insight lately. Think I missed something? Email me and tell me all about it.
It seems unlikely that lawmakers will be able to take any action on AI before it becomes entirely mainstream. Still, concerns arose this week when OpenAI CEO Sam Altman asked Congress for AI regulation “above a crucial threshold of capabilities.”
Via Axios, concerns that top the list include dangerous and harmful content, impersonation of public and private figures (remember Balenciaga Pope?), and, with 2024 already looming ahead of us: election misinformation.
Watch the whole testimony below while you’re clearing out your inbox (or writing an AI roundup that your boss now expects weekly).
After ending 2022 with barely a literal (virtual) leg to stand on, Meta has been dropping open-source AI innovations at a rapid pace in 2023. This week, they gave us a peek into the chip they’re using to power the Metaverse and generative AI technology. Meta dropped some juicy numbers behind their largest LLaMA model: LLaMA 65B contains 65 billion parameters and was trained on 1.4 trillion tokens.
The company also mentioned they developed an internal generative AI coding tool to help their developers work more efficiently, similar to GitHub Copilot.
Per CNBC, Meta has also been “overhauling its data center designs to focus more on energy-efficient techniques, such as liquid cooling, to reduce excess heat.” A theme — and an opportunity — we’ll see more of in the coming months is the incredible environmental toll of training AI models. Training one AI model is estimated to produce 626,00 pounds of carbon dioxide equivalent, “nearly five times the lifetime emissions of an average American car.”
What do investors want in an AI-driven business? Via Bloomberg, Insight’s very own George Mathew gave his take on what separates the hype from the real opportunity: the ability to have private data available to build AI products, a focus on user experiences, and having a great workflow that fits the way software fits within the industry. The combination of those aspects can make for a “very compelling” business.
Bits and bots
There’s an app for that. OpenAI dropped a ChatGPT iOS app this week, killing dozens (hundreds?) of third-party apps in the process. I’m sure Google is thrilled.
Meanwhile, the chatbot horserace continues. It seems the recent upgrade to PaLM 2 has given Bard a competitive advantage in head-to-head comparison.
Google Bard just got a massive upgrade.
ChatGPT is now falling behind.
Here are 10 things you can do with Bard that aren't possible with ChatGPT:
— Moritz Kremb (@moritzkremb) May 11, 2023
AI long-form. Insight’s AI chat went on a tangent this week when asked for their key reads on, “the impact AI could have on society and civilization.” Here are a couple of recommendations, if you’re looking for a good read during the upcoming Memorial Day Weekend:
- Of God and Machines
- Superintelligence by Nick Bostrom
- The Most Human Human and The Alignment Problem, both by Brian Christian (these are next on my list)
Weekend listening. Yann LeCun on Why Artificial Intelligence Will Not Dominate Humanity, Why No Economists Believe All Jobs Will Be Replaced by AI, Why the Size of Models Matters Less and Less & Why Open Models Beat Closed Models
Cybersecurity is going to get a lot more difficult. And I *just* successfully explained what phishing is to my mom.
Cybersecurity is about to get rough.
As suspected, AI is great at creating spearphishing attacks, emails that feel real but compromise security.
— Ethan Mollick (@emollick) May 16, 2023
The internet can’t get enough of the Wes Anderson aesthetic. From TikTok trends to incredible AI-generated movie trailers like my personal favorite below (happy belated May 4th).
Editor’s note: Articles are sourced from an ongoing, internal Insight AI/Data Teams chat discussion and curated, written, and editorialized by Insight’s VP of Content and Thought Leadership, Jen Jordan, a real human. (Though maybe not for long?)
Image credit: Google Deepmind via Unsplash. “Neuroscience.” Artist: Chris Schramm