Jake and Stake🚀🔎 (@JakeAndStake) header
Jake and Stake🚀🔎 (@JakeAndStake) profile picture

✍️ Learning in public 🚀 Infinite frontier 🔎 Infinite search

Free Newsletter ➡️
Joined June 2018
Photo by Jake and Stake🚀🔎 (@JakeAndStake)Photo by Jake and Stake🚀🔎 (@JakeAndStake)Photo by Jake and Stake🚀🔎 (@JakeAndStake)Photo by Jake and Stake🚀🔎 (@JakeAndStake)Photo by Jake and Stake🚀🔎 (@JakeAndStake)Photo by Jake and Stake🚀🔎 (@JakeAndStake)Photo by Jake and Stake🚀🔎 (@JakeAndStake)Photo by Jake and Stake🚀🔎 (@JakeAndStake)Photo by Jake and Stake🚀🔎 (@JakeAndStake)
The only metric that matters: The Power User Curve One metric tells you whether you have Product-Market Fit or a slow painful death
2
1
1
2
242
Meta's Product Integration Plus, AI generated ads will be good for their value-proposition The limit set by media departments will allow unlimited A/B testing of copywriting and images/video Meta is looking to use AI themselves in several ways: 1. Instagram pitched influencers on its AI chatbot that responds to fan comments and messages xitter.gq/JakeAndStake/sta… 2. AI features for advertisers to generate ad copy and background images. And a tool to fit their content for different aspect ratios required for various products, like the FB Newsfeed or Reels
🚨Instagram to replace influencer personalities with AI🚨 Instagram pitched influencers on its AI chatbot that responds to fan comments and messages Influencers are hesitant, "What if it says something that gets me cancelled? I should get cancelled on my own terms" 0/6
2
72
Open Source AI is Icing on the Cake Any open source models are just icing for Meta an Apple Meta already has the customer touch points needed to distribute any AI products And any AI generated content will be consumed on their properties: Facebook, Instagram, WhatsApp Open source AI models will be run on Apple silicon—as long as they have the best consumer hardware I'm mainly talking inference here. Leave training to the hyperscalers
1
1
103
Llama Takes the Throne The open source landscape has fierce competition. It seems like a new model comes out every few weeks topping the leaderboard Llama 3 8B and 70B outperforming similarly-sized models like Google's Gemma, Mistral 7B, and Anthropic's Claude 3 Sonnet The model has improved on Llama 2 wrt coding benchmarks The models were trained on a dataset 7x larger than Llama 2, containing 15T tokens and 4x more code
1
50
Back to the announcement Two smaller Llama 3 models have already been released, both in the Meta AI assistant and to outside developers While a much larger, multimodal version will arrive in the coming months
1
26
2. They're behind If Meta had a GPT-4 level model, they wouldn't open source it. They'd do what everyone else does Wrap it with chat interface and charge users for the privilege
1
10
1. AI is not their core business Meta makes money from serving ads alongside user-generated content More AI --> more content --> more ads --> more profits AI lowers the barriers to making content iPhone ➡️ everyone's a photographer AI ➡️ everyone's a photographer, editor, graphic designer, etc. It's never been cheaper to be a content creator
1
22
Why is Meta open sourcing their models? 1. AI is not their core business 2. They're behind Let's dig into their open source strategy in light of the newly released Llama 3 models • 8 billion parameters • 70 billion parameters
1
154
Google is healing
Google's CEO tells his employees to stop bringing their politics to work and respect coworkers right to just do their job. Let's see if this causes the same media and Twitter meltdown it did when we asked the same. I doubt it! blog.google/inside-google/co…
1
92
GenAI—content production—is a complement to Meta's advertising properties If another company controls it, they get margin on every prompt, not Meta If Meta makes it free, they benefit from more content
35
OpenAI makes the AI content more expensive because they charge to use their models They need AI to make money. Meta doesn't More expensive content --> more costly to share content --> less content --> etc
1
1
48
AI-made stuff will be shared on Meta, which is good for them. More content --> more eyeballs --> more ads --> more money It behooves Meta to let as many people as possible create more content with it If models are controlled by OpenAI/etc, the margin goes to them
I have a really stupid question: Why is Zuck spending so much on gen AI? What’s his long term benefit from this?
1
115
If you like this, please ❤️ Drop a like 🔄 Share with a friend 👤 Follow @JakeAndStake TY :) xitter.gq/JakeAndStake/status/17…
🚨Instagram to replace influencer personalities with AI🚨 Instagram pitched influencers on its AI chatbot that responds to fan comments and messages Influencers are hesitant, "What if it says something that gets me cancelled? I should get cancelled on my own terms" 0/6
36
Takeaways • IMO AI's still not good enough to do this • As it gets better it will create a new market • Which makes it cheaper for small influencers to engage fans • No matter what happens, Meta is still a big winner. AI content will be consumed on Meta properties 6/6
1
30
Many PR firms have started to use AI to generate responses on social media Meta will probably make a market for smaller creators and PR firms will continue to support larger ones This creates a market segmentation: high vs low end 5/6
1
47
Influencers want to • minimize cost of engaging with users • maximize the value of the parasocial relationship In sum-->reduce costs without damaging their reputation Most hire PR firms to handle their social media accounts, but Meta would be cheaper due to scale 4/6
1
9