Video · 2026-07-09 · 1h 8m · 24 moments

The rise of taste, human authenticity and judgment in an AI world | Adam Mosseri (Head of IG)

✦ AI generated

timeline · colored by role

Claim◆

Taste matters more, not less, in a world where AI makes building easy — the winners will be the ones who are clear-eyed about what AI is and isn't good at.

Mosseri argues that as AI makes execution cheap, the scarce and valuable skill becomes deciding what's worth building, and that requires taste and an instinct for AI's strengths and limits.

transcript

Adam Mosseri: No, I think taste matters a ton. In a world where it's easier to build things, it's more important to make sure that your time is spent figuring out what you should be building in the first place. The people who I think are going to make the most of it are the ones who are cleareyed about what AI is good at and what it's not good at and also have an instinct or a nose for what it will be good at and not good at.

extends · 1

Fact

Instagram is shifting from large teams of narrow specialists to small 'pod' teams of four to six generalist engineers plus a 'product staff' role that blends PM, design, and data science functions, calling in senior specialists only when truly needed.

Mosseri describes a major 2026 shift at Instagram/Meta: instead of a dozen narrow specialists, teams now consist of a handful of generalist engineers and a 'product staff' hybrid role, pulling in deep specialists only for specific needs.

transcript

Adam Mosseri: This year it's changing—we've adopted what we call pods, which are just mini teams where it's call it four to six engineers who are a bit more generalists. One we call product staff, which is sort of an evolution of the PM—a PM who can do some of what a designer does and some of what a data scientist does and some of what a researcher does.

Mechanism

Instagram is restructuring around small, generalist 'pods' of four to six engineers plus a 'product staff' role that absorbs former PM, design, and data-science work, calling in specialists only when truly needed.

Mosseri describes Instagram's 2026 shift from baker's-dozen functional teams to lean pods led by a generalist 'product staff' role, with specialists brought in only for specific needs.

transcript

Adam Mosseri: This year it's changing, we've adopted what we call pods which are just mini teams where it's call it four to six engineers who are a bit more generalists... So we try to build the team based on the needs of the work a bit, but then end up with a much smaller core which is more on the order of six or seven usually.

gives example · 1

Mechanism

Instagram has replaced large teams of narrow specialists with small four-to-six person generalist 'pods' centered on a new hybrid 'product staff' role, pulling in senior specialists only when the work truly requires them.

Meta/Instagram is restructuring teams from baker's-dozen specialist squads into smaller 'pods' built around generalist 'product staff' who absorb design, data science, and research work using AI tools.

transcript

Adam Mosseri: we've adopted what we call pods which are just mini teams where it's call it four to six engineers who are a bit more generalists. Uh one we call product staff which is sort of an evolution of the PM. So a PM who can do some of what a designer does and some of what a data scientist does and some of what a research does leveraging um the latest tools that we have for them.

gives example · 1supports · 1

Fact

Instagram has shifted this year to small, generalist 'pod' teams of four to six engineers plus a 'product staff' role that blends PM, design, and data science work, replacing the older, larger team of specialized functional roles.

Adam describes Instagram's move from baker's-dozen specialist teams to small 'pods' with a generalist 'product staff' role that absorbs work formerly done by dedicated designers, data scientists, and researchers.

transcript

Adam Mosseri: This year it's changing we've adopted what we call pods which are just mini teams where it's call it four to six engineers who are a bit more generalists. Uh one we call product staff which is sort of an evolution of the PM. So a PM who can do some of what a designer does and some of what a data scientist does and some of what a research does.

Claim◆

In a world where AI makes building things easy, taste — the judgment of what to build in the first place — becomes the scarce and valuable skill, which is why designers who have taste are hard to automate away.

Mosseri argues that as AI removes friction from building, the bottleneck shifts to deciding what's worth building — which is why he remains bullish on designers and taste-driven roles despite industry anxiety.

transcript

Adam Mosseri: In a world where it's easier to build things, it's more important to make sure that your time is spent figuring out what you should be building in the first place... I'm actually pretty long on design or designers because they tend to have taste and I think that is something that is much more difficult to imagine being automated away.

extends · 1

Claim◆

In a world where it's easier to build things, taste — knowing what to build in the first place — becomes more valuable, and designers are undervalued right now precisely because they possess that hard-to-automate taste.

Adam argues that as AI makes execution cheap, the scarce and valuable skill becomes taste — knowing what's worth building — which is why he's bullish on designers despite widespread anxiety about their roles.

transcript

Adam Mosseri: No, I think taste matters a ton. Uh, so in a world where it's easier to build things, it's more important to make sure that your time is spent figuring out what you should be building in the first place. Um, actually, so a lot of designers right now are very anxious about their roles... but I'm actually pretty long on design or designers because they tend to have taste.

extends · 1

Claim◆

Taste—knowing what to build and what AI is and isn't good at—matters enormously now, and that's precisely why designers remain highly valuable even as generalists take over more of the design work.

Mosseri argues that as building things gets easier with AI, the scarce and valuable skill becomes knowing what's worth building—which is why he remains bullish on designers, since taste is hard to automate.

transcript

Adam Mosseri: No, I think taste matters a ton. So in a world where it's easier to build things, it's more important to make sure that your time is spent figuring out what you should be building in the first place. I'm actually pretty long on design or designers because they tend to have taste and I think that is something that is much more difficult to imagine being automated away.

provides context · 1

Claim

As AI absorbs more of the product development lifecycle, the enduring human contribution shifts toward vision, strategy, and judgment rather than execution, since people give AI bounded goals and constraints rather than asking it to originate strategy outright.

Mosseri says the durable human value in product work is taste and judgment — especially setting vision and strategy — since people direct AI within constraints rather than asking it to invent strategy from scratch.

transcript

Adam Mosseri: taste like we talked about um judgment particularly around strategy right like you're you're not you might get feedback from an AI on a strategy but you're not asking an AI to come up with a strategy anytime soon or if you are then it's within the context of bounds you set so here's my goal here's my vision here are my constraints here's my job here's my budget.

explains mechanism · 1extends · 1gives example · 1provides context · 1

Claim

As AI eats more of the product development life cycle, human value concentrates in taste, judgment, vision, and strategy rather than execution.

Mosseri says humans won't hand strategy creation to AI outright; instead their role shifts toward defining vision, setting bounds, and giving feedback, much like managing a team.

transcript

Adam Mosseri: Taste like we talked about, judgment particularly around strategy—you might get feedback from an AI on a strategy but you're not asking an AI to come up with a strategy anytime soon. I think of vision as an articulation of the world or the state of the product you want to get to. Then I think of strategy as an opinionated path to achieve that vision.

extends · 4gives example · 2

Claim

As AI eats more of the product development lifecycle, human value concentrates in vision and strategy — defining the goal and an opinionated, deliberately contestable path to reach it — rather than in execution.

Mosseri argues that taste and judgment — especially strategy, defined as a deliberately controversial, opinionated path to a vision — will remain the domain where human brains add the most value as AI takes over execution.

transcript

Adam Mosseri: Taste like we talked about, judgment particularly around strategy... I think of vision as an articulation of the world or the state of the product you want to get to. Then I think of strategy as an opinionated path to achieve that vision. If a strategy can't be like 'be the best' or 'be amazing,' it has to be controversial — a reasonable person should be able to disagree with it.

Claim◆

As AI eats more of the product development lifecycle, human value will concentrate in vision and strategy rather than execution — because strategy requires weighing messy, non-obvious inputs like team dynamics and competitive positioning, and AI only produces good strategy when heavily steered by a human.

Adam argues human brains remain most valuable for vision and strategy — defining goals, constraints, and judgment calls — because AI strategy output is only good when a human steers it with the right context and constraints.

transcript

Adam Mosseri: Taste like we talked about, judgment particularly around strategy — you might get feedback from an AI on a strategy but you're not asking an AI to come up with a strategy anytime soon... I think of vision as an articulation of the world or the state of the product you want to get to. Then I think of strategy as an opinionated path to achieve that vision.

Claim

AI does not naturally produce good strategy on its own; it only becomes useful for strategy when a human aggressively steers it with real constraints like technology state, team dynamics, competitive and regulatory landscape, and brand identity.

Despite having access to market data, AI produces only predictable, generic strategy unless a human deliberately steers it with the full set of real constraints — team, competition, regulation, and brand identity.

transcript

Adam Mosseri: I think it could be. I I have found it's not unless you steer it pretty aggressively. And I don't mean towards an answer. I mean based on the constraints. It turns out when you're trying to come up with a strategy, there's a lot of things to consider, right? You need to consider the state of the technology, the personnel on the team and what's motivating them and what you can get.

Fact

The common assumption that Instagram's algorithm has a detailed semantic understanding of each user's specific interests is backwards — for years it has run mainly on illegible embedding vectors rather than legible concepts like 'likes surfing,' and only recently has that become more semantically readable via LLMs.

Mosseri corrects a widespread misconception: Instagram's ranking system historically worked on unreadable embedding vectors, not human-legible concepts about a user's interests, and only recently has LLM technology started making those signals describable in plain language.

transcript

Adam Mosseri: One of the most common misconceptions is actually in the opposite direction. I think people assume that there's a much more detailed semantic understanding of everybody's interests and preferences in the algorithm than there is. Most of what's really driven the progress in the world of recommenders over the last 5 10 years have been, you know, these large embedding models and these other techniques that basically produce artifacts that cannot be read by people.

Mechanism

The Instagram recommendation algorithm does not have a detailed semantic understanding of users' interests; it operates on illegible embedding vectors, not concepts like 'likes surfing.'

Mosseri corrects a common misconception: rather than 'knowing' that a user likes surfing, the algorithm relies on massive, non-human-readable embedding vectors, and only now with LLMs can that be translated into semantic descriptions.

transcript

Adam Mosseri: One of the most common misconceptions is actually in the opposite direction. I think people assume that there's a much more detailed semantic understanding of everybody's interests and preferences in the algorithm than there is. Most of what's really driven the progress in the world of recommenders... have been these large embedding models... that basically produce artifacts that cannot be read by people.

Fact◆

Instagram's recommendation algorithm does not have a rich, human-legible semantic understanding of users' interests; it relies on large embedding models producing illegible high-dimensional vectors, and only recently have LLMs made it possible to translate those vectors into human-readable descriptions.

Adam corrects a common misconception: the algorithm doesn't 'know' users like a person would — it works off illegible embedding vectors, and LLMs are only now enabling those to be translated into human-readable interest descriptions.

transcript

Adam Mosseri: I think people assume that there's a much more detailed semantic understanding of everybody's interests and preferences in the algorithm than there is. Most of what's really driven the progress in the world of recommenders over the last 5, 10 years have been, you know, these large embedding models... they're not legible. They're like giant vectors.

Fact

The common belief that Instagram's algorithm has a detailed, human-readable understanding of each user's interests is wrong — it actually relies on illegible high-dimensional embedding vectors, not semantic concepts like "likes surfing."

Mosseri says people overestimate how much the algorithm 'understands' them semantically — under the hood it's just correlated numeric embeddings, not legible concepts, though LLMs are starting to make those embeddings describable.

transcript

Adam Mosseri: I think people assume that there's a much more detailed semantic understanding of everybody's interests and preferences in the algorithm than there is... It knows I like surfing and it's like it doesn't. It just has this big ass number that happens to correlate with surfing.

Mechanism

Pure chronological feeds break down at scale because they incentivize overposting by professional accounts, drowning out friends, and they ignore that recency isn't the only signal of relevance.

Mosseri explains why Instagram doesn't default to chronological feeds: at scale it rewards high-volume posters like publishers over friends, and testing shows overall satisfaction drops when feeds go chronological.

transcript

Adam Mosseri: There are a couple issues with the chronological feed. So one is... the tension between an individual's interests and what works when you scale it up. If you do a pure chronological feed, the incentive for everybody is to just post as much as possible because it will always be at the top of everyone who follows you's feed as soon as you post.

Mechanism

A pure chronological feed doesn't actually serve users better because it creates a perverse incentive for accounts to overpost, letting high-volume publishers and professional content crowd out infrequent posts from close friends and family.

Adam explains why chronological feeds fail at scale: they incentivize overposting by publishers and large accounts, drowning out infrequent posts from friends, and testing shows satisfaction actually drops when chronological is the default.

transcript

Adam Mosseri: If you do a pure chronological feed, the incentive for everybody is to just post as much as possible because it will always be at the top of everyone who follows you's feed as soon as you post. So what ends up happening is that the feed gets overwhelmed with professional content... your best friend won't — you might get one thing a week from them — and so your feed just gets taken over.

Mechanism

A pure chronological feed fails at scale because it creates incentives for high-volume professional posters to dominate everyone's feed, crowding out less frequent but more meaningful posts from friends and family, and because recency is only one input into relevance, not the only one.

Mosseri explains why chronological feeds break down at scale: they reward posting volume over relevance, letting publishers flood feeds while rare-but-important posts (like a sister's engagement) get buried, and testing shows satisfaction actually drops under chronological defaults.

transcript

Adam Mosseri: If you do a pure chronological feed, the incentive for everybody is to just post as much as possible because it will always be at the top of everyone who follows you's feed as soon as you post... your best friend won't — you might get one thing a week from them — and so your feed just gets taken over. Recency is an important input into relevance, but it's not the only one.

Prediction◆

The rise of AI-generated content will be a net tailwind for Instagram because in a world flooded with synthetic content, people will increasingly seek out authentic human creators and points of view rather than avoid AI content altogether.

Mosseri predicts AI content is a tailwind but a challenge for Instagram, since abundant synthetic content will make people crave authenticity and real creators, an area he thinks Instagram is best positioned to serve.

transcript

Adam Mosseri: I think it's going to be a tailwind, but I think it's going to be a challenge... In a world where there's an abundance of synthetic content, I actually think people are going to seek out creativity and authenticity and people more, not less. I don't think we should filter out AI content. I think we should let you know if content is AI content or not.

Prediction◆

The rise of AI-generated content will be a net tailwind for Instagram, not a headwind, because in a world flooded with synthetic content people will increasingly seek out authenticity and real people, which is Instagram's core strength as the largest creator platform.

Mosseri predicts AI content will help rather than hurt Instagram: as synthetic content proliferates, users will crave creativity, authenticity, and real people behind the content—an area where Instagram, as the largest creator platform, is well positioned.

transcript

Adam Mosseri: I think it's going to be a tailwind, but I think it's going to be a challenge... I don't think we're very good at ranking AI content yet. There's great AI content. There's crap AI content. In a world where there's an abundance of synthetic content, I actually think people are going to seek out creativity and authenticity and people more, not less.

Prediction◆

The rise of AI-generated content will ultimately be a tailwind for Instagram because, amid an abundance of synthetic content, people will increasingly seek out creativity, authenticity, and real people — an area where Instagram's investment in individual creators positions it well.

Adam predicts AI content will be a net tailwind for Instagram despite ranking challenges, because abundant synthetic content will make people crave authentic, human creators more, not less — an audience Instagram has long invested in.

transcript

Prediction◆

As synthetic AI content becomes abundant, people will seek out creativity, authenticity, and real people more rather than less, which makes the rise of AI content a net tailwind for Instagram's creator-driven platform, so long as content isn't judged by which tool made it.

Mosseri predicts that a flood of synthetic AI content will increase, not decrease, demand for authentic human creators, positioning Instagram's creator focus as a tailwind rather than a threat from AI content.

transcript

Adam Mosseri: In a world where there's an abundance of synthetic content, I actually think people are going to seek out creativity and authenticity and people more, not less. And I think that that will help us. That doesn't mean that we won't have AI content on our platform. There's going to be bad and good AI content and we're going to try and handle that, you know, the way we normally handle content.

supports · 1

Highlight slides

Related episodes

Video · a16z Podcast

Ben Horowitz on the Global Race for Tech, Power, and Influence

→

Audio · All-In Podcast

SpaceX-Cursor Deal, SaaS Debt Bomb, New Apple CEO, SPLC Indictment, Colon Cancer Spike

→

Video · a16z Podcast

Software in the Age of Agents | The a16z Show

→