00:00Precision delivery of medicine
00:02entertainment franchise games absolutely
00:05exploding small modul reactors and the
00:07nuclear Renaissance plus AI moving into
00:11very complex workflows now these were
00:14just a few of the major Tech innovations
00:16that Partners at a16z predicted last
00:19year and our partners are back we just
00:21dropped our list of over 40 plus big
00:24ideas for 2024 a compilation of critical
00:27advancements across all our verticals
00:30from Smart energy grids to Crime
00:31detecting computer vision to
00:33democratizing Miracle drugs like gp1s or
00:36even AI moving from blackbox to clear
00:40boox you can find the full list of 40
00:42plus Builder worthy Pursuits at a16
00:46z.com bigideas 2024 or you can click the
00:50link in our description below but on
00:53Deck today you will hear directly from
00:55one of our partners as we dive even more
00:57deeply into their big idea what what's
01:00the why now what opportunities and what
01:02challenges are on the horizon and how
01:04can you get involved let's dive
01:07in as a reminder the content here is for
01:10informational purposes only should not
01:12be taken as legal business tax or
01:14investment advice or be used to evaluate
01:16any investment or security and is not
01:18directed at any investors or potential
01:20investors in any a16z fund please note
01:24that a16z and its Affiliates may also
01:26maintain investments in the companies
01:27discussed in this podcast for more
01:30details including a link to our
01:31investments please see
01:38disclosures hi everyone I'm Jonathan Li
01:41I'm a general partner on a6c games and
01:46my big idea for 2024 is AI first schemes
01:50that never end so in 2024 I believe
01:54we'll see the first cohort of AI first
01:56games from creators that by which large
01:59language models to enable novel game
02:01systems and mechanics and while much of
02:04the early discourse on generative Ai and
02:06games is focused on how AI can make game
02:08creators more efficient I actually
02:11believe the largest opportunity
02:13longterm is in leveraging AI to reinvent
02:16the nature of the games themselves um so
02:19creating NeverEnding games that engage
02:21and retain users for a very very long
02:24time um a couple of examples of this uh
02:27generative agents powered by large
02:30can create incredibly lifelike
02:32companions in emergent social behavior
02:34um toble charging games that heavily
02:37utilize non-player characters or or NPCs
02:41um personalized character Builders and
02:43narrative systems will enable every
02:44player to have a fully unique and
02:47personalized playthrough of their
02:48favorite game um and in the game worlds
02:51themselves can no longer will no longer
02:54be rendered but actually generated at
02:56runtime using neural networks a new
02:59player on boarding itself is poised to
03:02reinvented um and then my idea is that
03:05every game is potentially will
03:07potentially be redesigned around an AI
03:09co-pilot with the Mantra of good alone
03:12great with AI and best for friends
03:15Jonathan you mentioned several
03:16opportunities or even you could say
03:18fundamental shifts in game development
03:20from never- ending worlds to convincing
03:23NPCs to personalized real-time rendering
03:26I want to know which of these upgrades
03:27you are already seeing built perhaps and
03:30also which you're most excited about yes
03:33so we generally think of the AI
03:34revolution in games is happening in two
03:37waves the first wave is building the
03:39same game but faster and higher quality
03:42with AI tools and infrastructure to
03:44empower game developers and the exciting
03:47thing is that this first wave is already
03:49happening right now it's not a theory
03:52it's reality so we surveyed 230 game
03:55studios earlier this year and it turns
03:58out that 87% of them are already using
04:01generative AI tools and so some of the
04:03products that they named include chat
04:05GPT mid Journey 11 Labs K kadm Adobe
04:09Firefly GitHub co-pilot Etc and the
04:12majority of these students rated sorry
04:14the majority of these Studios rated the
04:16impact of these tools as
04:17transformational which is pretty
04:20incredible the second wave uh which we
04:23consider even more exciting it's
04:25building entirely new sets of AI native
04:28games that are uniquely enabled by Ai
04:31and those are the examples I write about
04:33in my big idea for 2024 and so Dynamic
04:36worlds that can be generated in real
04:38time based upon a player's actions you
04:40know lifelike AI companions never ending
04:43to the Choose Your Own Adventure Games
04:45we still very early in seeing what those
04:47new AI native game game experiences
04:49might be like and so you've seen a
04:51couple of prototypes some some
04:53interesting sort of proof of Concepts
04:55but in general the the plane field is
04:58wide open and that's the exciting thing
05:00um for the folks that do get it right
05:03like these are these are the ideas that
05:06can potentially create entirely new
05:08genres to Captivate you know hundreds of
05:10millions of people and that's what we're
05:11incredibly excited about looking out in
05:13the 2024 yeah I think I think what
05:16you're picturing is is really important
05:18because what you're saying is is we're
05:21just incrementally optimizing existing
05:25games you're really describing a fully
05:27new game environment fully new way of
05:29thinking about games and I love this
05:31Mantra that you included which is good
05:32alone great with AI and best with
05:35friends can you elaborate on what that
05:38means and maybe how it shifts or has the
05:40potential to shift the way that people
05:42currently think about both building
05:44games and also playing games so many of
05:47the biggest games in the world today you
05:49think about fortnite Call of Duty League
05:51of Legends they utilize AI primarily for
05:54onboarding and practice and so before
05:56you play against real human opponents
05:59you might play against AI Bots and they
06:01enable you to practice build confidence
06:03learn all of the controls get your hand
06:05eye coordination down and so it's sort
06:07of a warm-up before you jump into like
06:10unquote I think one of the promises of
06:13generative agents is in elevating agents
06:16like the AI Agents from a warmup into
06:18the core game itself and so one example
06:21of that that I find fascinating this
06:23this concept of an AI companion for
06:25every game so let's say at the beginning
06:27of a game like you're starting Call of
06:29Duty for example you know you might
06:31customize your character and let's say
06:33you also customize an AI companion that
06:36you can bring with you into the game and
06:39this companion can play the entire game
06:41with you right like either as um on on
06:45the same team or against you if needed
06:47and it might comment on the experience
06:49it understands the world around it it
06:51keeps you company um when your friends
06:53aren't there to play with you it learns
06:55from all of your preferences over time
06:57and helps you out and generally tries to
06:59best to make sure that you have a great
07:00time and so such an agent wouldn't
07:04replace pling of humans but it would
07:05certainly be better than playing by
07:07yourself and so that's what I mean by
07:10designing games that are good alone they
07:12can be played single player they can
07:13also be played with an AI and it's great
07:15with AI but of course it's best if you
07:18have human friends to play with and
07:21those different modalities can be mixed
07:22and matched right depending on the game
07:24and depending on the time yes what gives
07:26you the confidence though that the these
07:29are the type of games when we're talking
07:31about the inclusion of AI becoming
07:33hyper-personalized potentially what
07:35gives you the confidence that those are
07:37the type of games that people want and
07:40I'm asking that as it relates
07:41specifically to community and the fact
07:43that games really are these communal uh
07:46interactive environments where people
07:48have somewhat of a shared reality and so
07:51if everything is hyper-personalized I'm
07:53curious to know your perspective about
07:55how that shifts people's ability to
07:58build community and what are you seeing
07:59in terms of you mentioned some of this
08:01is already being implemented what are
08:03you seeing that gives you the indication
08:05that this hyper-personalization or
08:07inclusion of AI is something that people
08:10want so the interesting thing about um
08:13games is that uh the games industry has
08:15actually embraced AI for decades and so
08:18in a way the technology that the latest
08:20expression of the technology large
08:22language models is new but the consumer
08:25Behavior the patterns around it are
08:27actually um already existing and so that
08:30that's what gives me confidence that
08:31there's demand for some of these um new
08:33types of gameplay that we talk about so
08:36take AI companions for example um it
08:38might sound like a crazy idea at first
08:40flush but so many iconic games they
08:43already have ai companions that are
08:45prescripted and they don't use large
08:47language models but players have still
08:50developed deep deep attachments with
08:51them a great example is the Halo
08:53franchise there's an AI companion in
08:55that game called quana and she talks to
08:58Master Chief comments and what he's
09:00doing basically like is a it's a chip
09:02that rides in his helmet and he and she
09:04tells him where to go next when he's
09:06lost and is there from the beginning to
09:08the end of like almost every Halo game
09:11and there are millions of people all
09:13over the world um that adore Cortana
09:15right fanfiction about you know Cortana
09:18in the Halo Universe um to the point
09:20where I believe Microsoft has integrated
09:22Cortana into its that core Windows
09:24search function um and in another game a
09:28favorite of mine The Last of Us which is
09:30a famous Naughty Dog franchise that was
09:32recently turned into an HBO share you
09:34play as Joel but then you have Ellie as
09:37an AI companion and she follows you
09:39around for basically the entire game she
09:41adapts to your behavior if you're
09:43sneaking somewhere she'll also sneak
09:44with you if you're just in a firefight
09:46she helps you out in combat situations
09:49and she has a fully fleshed out
09:50personality that you learn more about
09:52over the course of the game um and so
09:55from that perspective AI companions have
09:57already existed and are very very
09:59popular right and so the idea of just
10:02toble charging this of large language
10:03models to make them more lifelike and
10:05enable them to have a wider range of
10:07expression and behavior I think is going
10:09to be very well accepted by players on
10:13hyper personalization which is the um
10:15the question that you were asking around
10:17um storytelling and and how that works
10:20of communities I'm optimistic based on
10:23the fact that EV that personalization
10:26already happens to a large degree in
10:29uh role playing games like balers Gate 3
10:32have incredibly detailed character
10:33Builders where you can customize
10:35everything from your skin color did the
10:37highlights in your hair did the tattoos
10:39in your face and people spend hours
10:41customizing their avatars right like
10:43their entire YouTube channels dedicated
10:45to how to create the most like lifelike
10:47or the most badass um character for for
10:50each of these games and let's not forget
10:52that a huge portion of the revenue in
10:55games like fortnite and League of
10:56Legends comes from selling skins which
10:59are you know personalized virtual
11:01clothing for how your character might
11:03look in game one of the oldest RPGs of
11:05all time is Dungeons and Dragons which
11:07by definition is a personalized
11:10Adventure that is told by a human
11:12narrator a dungeon master a game master
11:15he basically crafts an improv adventure
11:17with you and your group of friends right
11:19and so these are just two examples of um
11:22you know personalization in games and an
11:24AI companion in games but I'm I'm
11:26optimistic that players have already
11:28embraced many of the concepts that
11:30generative AI is turble charging and so
11:33from that perspective we're not in we're
11:34not inventing entirely new consumer
11:36Behavior we're just making it easier for
11:39players to do what they already want
11:41yeah absolutely and you mentioned a few
11:44you could say age-old games Halo
11:46Dungeons and Dragons they've been around
11:48for quite some time do you see this new
11:51technology or is that you said a
11:52technology that has been ingrained in
11:54games for also um some time do you see
11:57that really benefiting the incumbents
11:58who already have these large player
12:00bases or do you see this kind of
12:02fundamental shift where you're going to
12:04see completely new winners in this
12:06completely new era how do you think
12:08about maybe some of the new entrance
12:10versus the incumbents and and who really
12:12wins so I think the incumbents let's
12:14call them sort of that the big triaa
12:16game developers and Publishers um they
12:19are already going deep into the wave one
12:22AI tools and infrastructure um that I
12:25spoke about and so I think almost all of
12:27them are using chat gbt mid Journey
12:29Adobe Firefly um bunch of these uh
12:32southw tours earlier and I think they
12:34will benefit hugely from them as as
12:36still enable them to make higher quality
12:38games faster get more shots and goal and
12:40and so on I am skeptical however that
12:43the largest developers will be the ones
12:45that create incredibly Innovative AI
12:47native games of the sort that we we
12:50spoke about as as the wave PE games um
12:53and many reasons but one of them is that
12:56if you're a large developer your
12:58business revolves around IP so you
13:01typically have a few key franchises you
13:03know Halo Dungeons and Dragons World of
13:05Warcraft Call of Duty and your business
13:07revolves around launching sequels to
13:10those franchises every few years and so
13:12you don't actually need to take that
13:14much risk and and shipping these sequels
13:17but in most cases the players themselves
13:19actually aren't looking for something
13:21too different um like the next Call of
13:23Duty title is like comfort food you know
13:25exactly what you're going to get and
13:27that's why you enjoy it um yeah so I
13:30think the business model of these large
13:31incumbents I think doesn't support
13:33taking too much risk um so I think
13:36that's one reason and
13:39second I posit that the first generation
13:41of these AI native games might end up
13:43actually looking really silly um to
13:46traditional game developers and so they
13:47might not actually take them seriously
13:49at first um and so take mobile games for
13:52example which was um uh the last sort of
13:54major sort of Paradigm shiting games the
13:57winners in mobile games were largely new
13:59companies uh supercell keen zingga and
14:03not the PC consoling companies right it
14:06wasn't Electronic Arts or Activision or
14:08Microsoft um and I bet a large reason
14:11for this was that if you were an
14:12executive producer at Activision and you
14:15looked at the game like Candy Crush in
14:16the beginning you probably thought hey
14:18this is like really silly right like
14:20Graphics are really bad like you know
14:23there's no multiplayer on this thing
14:24right like um it's a really really
14:27simple puzzle game so how how can this
14:29possibly be good right like how many
14:31people are play this thing um but what
14:34what I think they underestimated is that
14:35with each new paradigm of games a new
14:38set of players also comes into the
14:40ecosystem so the folks that played Candy
14:42Crush were for the most part older women
14:45and it was an entirely different
14:46demographic from the young men that
14:48played Call of Duty for example right um
14:51but the N effect was that it expanded
14:53the total high for games like the gaming
14:55time increased and over the course of
14:57that it created several valuable
14:59companies um another more recent
15:02examples um take a product like
15:05character AI where you have millions of
15:07people to chat with like their favorite
15:11characters I would guess that the
15:13majority of the users of character AI
15:15don't describe it as a game but you
15:18could you could squint and say hey like
15:20this is a consumer experience that is
15:22just as engaging as playing the game um
15:25but it's probably not on the radar of
15:27any of the traditional game developer
15:29was the incumbents that we just talked
15:32right yeah so I think in in that same
15:35vein I think the um the most Innovative
15:37AI native experiences are are are more
15:40likely to be created by startups and
15:42teams that are just getting going and
15:43are bringing that fresh perspective as
15:45opposed to um uh coming at it with um uh
15:50with a lot of history and and baggage
15:52around existing IP that they need to
15:54Shepherd yeah I mean I think that's such
15:56a good point about just the thing things
15:58that are emerging that people just would
16:00not even describe as a game but if you
16:03really think about it are not so
16:05different fundamentally from from games
16:08maybe super quickly can we touch on the
16:10business model because as these things
16:12fundamentally fundamentally take new
16:14shapes and forms so to do the things
16:18that people are willing to pay for but
16:20also when we think about AI the cost
16:22structure is a little bit different too
16:24right if you're rendering things in real
16:25time that has a a real inherent cost to
16:28it and so maybe you could just speak
16:29quickly to how you might see the
16:31business models evolving just like we we
16:33saw with mobile and then also um yeah
16:36like maybe things that developers should
16:39keep in mind if they are building these
16:41new games and are considering completely
16:43new business models yeah so one of the
16:46challenges is that games as a business
16:48model have trended towards uh two models
16:50so one is free to play with a majority
16:53of users actually around 90 to 95% on
16:56average play a game for free
16:59and in the minority 5 to 10% actually
17:01monetize and so the free users entertain
17:04the monetizing users and so they pay
17:06with their time even if they don't spend
17:08money and so some of the biggest games
17:10in the world like League of Legends like
17:12fortnite um are a freeto playay games
17:16and and then the second business model
17:17is a premium uh game and so a game that
17:20is just sold for you know 20 30 40 50
17:24bucks you pay you pay for that game up
17:25front and then you never spend again
17:29out the challenge is that um with an AI
17:32game if you have a game that's running a
17:33large language model in the cloud the
17:35influence costs of serving users
17:38dynamically generated worlds or tax
17:40would probably make either of these
17:41models unprofitable right off the bat
17:43right um because you're monetizing your
17:46user right up front for 50 or 60 bucks
17:49and then you're not you know in a
17:50premium game sale and then you're not
17:52getting any more Revenue going forward
17:55or in a freet to-play model like 90% of
17:57the users are actually just a cost right
17:59because they're they're not actually
18:00spending any money in the game um my
18:04guess is that um companies like mid
18:07Journey or Char like like character AI
18:09have been able to address this with a
18:10subscription model and so basically
18:12every power user pays for their own sort
18:15of like inference cost so to speak and
18:17we could see that subscription model
18:19also become the default business model
18:21for AI games and I think the game moves
18:23would be okay with it as long as they
18:25perceived the value was there um and as
18:28an interesting sort of um side point
18:31subscriptions have actually become more
18:33popular in games over the last sort of
18:34like call it five years with the r of
18:37services like Xbox game pass Apple
18:39arcade and even in inside individual
18:41games like fortnite they now have
18:43seasoned subscriptions like the like the
18:45battle pass where you play for a certain
18:47period of time and you subscribe up
18:49front and it unlocks to the pokes over
18:51the course of the season um yeah so I so
18:55my guess is that subscriptions become
18:56more popular for for AI games yeah
19:00absolutely one just super quick
19:02follow-up question there is if people
19:04are more linked to their game play as in
19:08the time that they're spending in game
19:11as it relates to inference do you think
19:13that creates a negative feedback loop
19:15whereas right now you can kind of like
19:17you choose to play a game and you can
19:19play it for quite some time and really
19:21get ingrained in it and you could say
19:23obsessed or addicted to it um without
19:26additional cost do you see
19:29kind of adverse relationship there where
19:31all of a sudden if my inference is
19:33linearly tied to cost that I may not be
19:37as likely to you know get as deeply
19:40involved in a game yeah no I think
19:42that's that's a great point that games
19:44will need to design around um I think
19:47the good news is that that same
19:49trade-off of Time Versus money also
19:52currently exist in freet toop playay
19:53games so um uh if you recall the early
19:58days of um Facebook games um take for
20:00example fville like the Dr Zinger games
20:03yes they actually attempted to gate the
20:05amount of time that you spend in the
20:07game behind an energy meter and so every
20:09day you would get call it 20 energy and
20:12every action in the game would cost like
20:14two or three energy and then when your
20:16energy was was gone you you could either
20:19take a break and then come back and play
20:21again tomorrow that's that's how the
20:22term appointment gaming came about or
20:24you could monetize and you could
20:26actually spend to refill your energy
20:28meter and that that means that enables
20:30you to keep going right and so you can
20:32imagine a similar thing happening with
20:33with AI games where um you have some
20:37measure of the influence cost for
20:38players and that's approximated by a
20:40virtual currency inside the game um and
20:44so uh I think it's just something that
20:46that game developers will need to design
20:47around but I think that Paradigm of like
20:49hey let's balance time versus money I
20:52think already already exists right yeah
20:55that's a great point and you're bringing
20:57me back because I spent far too much
20:58time playing Farm Bill back in the day
21:01um but to close out clearly there's
21:04opportunity on the table and I think
21:05it's it's really does feel like there is
21:08this this new era potentially of games
21:12and that's exciting so with that
21:14opportunity where would you start if you
21:17know you're a builder in the ecosystem
21:19what would you be looking at in 2024
21:21yeah it's it's very exciting um I would
21:25uh I would suggest that if you are
21:27excited about games take your favorite
21:29game genre whether it's a sports game
21:32like an MMO even chess or poker or
21:35Dungeons and Dragons and try to
21:37reimagine on first principles what it
21:39could look like if you were to introduce
21:41generative AI into the core gameplay
21:43Loop and then see if you can hack
21:45together a prototype or team up with
21:46friends to hack hack something together
21:49um and then come on over and share it to
21:51us like we we love playing crazy games
21:54and and wild ideas and um the reality of
21:57the games industry is that some of the
21:59best games got started by solo
22:01developers or um as mods and passion
22:04projects that are built by small teams
22:06you know DOTA like pubg like um
22:10Counterstrike like some of the most
22:11iconic games were built by very very
22:13small teams and then um I think uh an
22:17even more exciting thing is that anytime
22:19you have a new paradigm shift like we
22:20talked about mobile um early on I think
22:24the playing field advantage goes to new
22:26teams the folks that are coming in the
22:28brush perspective um and uh my guess is
22:32that um the AI game teams that get it
22:35right might actually end up looking very
22:37different from the traditional game
22:39developers of today right and if you get
22:42it right I think the potential is you
22:44can disrupt a $200 billion Market in the
22:46fastest growing category of media today
22:48so it's very exciting and um we love to
22:52we'd love to talk to you all right I
22:54hope you enjoyed this big idea we do
22:56have a lot more on the way including
22:58programmable medicine that's taking a
22:59page out of the reusable rocket Playbook
23:02anime going mainstream and whether the
23:05consumer AI Battleground may be moving
23:07from model to ux plus if you want to see
23:10our full list of 40 plus Big Ideas right
23:13now you can head on over to
