2025 AI Prediction Scorecard
Key Points
- The speaker reviewed 17 tech predictions made in January 2025, using a self‑created grading rubric to assess which were accurate (“hits”) and which missed the mark.
- Seven predictions were deemed “home runs,” with the strongest being the rise of AI‑only creators who are now earning six‑figure incomes and prompting the emergence of AI‑native creative agencies.
- A major shift in social media content was confirmed: roughly 40 % of Instagram feeds now contain AI‑generated material, making it increasingly hard for users to differentiate synthetic from human‑created posts.
- AI integration is deepening across platforms, exemplified by tools like Google’s Genie 3 and Meta’s internal AI features, even leading to recursive AI systems where one model navigates worlds generated by another.
Sections
- 2025 Prediction Review: Hits & Misses - The speaker revisits their 17 2025 forecasts, explains their grading rubric, and highlights seven predictions that came true, notably the rise of openly AI‑generated creators.
- AI Voice Companions Go Mainstream - The discussion highlights emotional AI voice companions becoming mainstream, backlash over voice changes, rapid adoption of AI tools across non‑tech jobs and workplace platforms like Slack, and the emergence of visual AI companions driving a growing multi‑hundred‑million‑dollar market.
- AI Dating Trends Self‑Assessment - The speaker reviews their recent hits and partial successes, then predicts that AI‑powered virtual companions will expand in dating platforms despite social etiquette and ethical challenges.
- Reassessing AGI Timelines and Debate - The speaker reflects on delayed AGI expectations, the increasing oddities in AI performance, and a shift toward personal AI safety concerns.
- Review of Missed AI Forecasts - The speaker evaluates a series of optimistic AI predictions—home visual feedback tools, a robot “iPhone moment,” Super Bowl robot ads, laundry‑folding bots, and AI weddings—admitting most missed their mark, noting only companionship remains a viable prospect.
- Request for Grading Feedback - The speaker asks viewers to identify any missed points or grading mistakes and share their thoughts in the comments.
Full Transcript
# 2025 AI Prediction Scorecard **Source:** [https://www.youtube.com/watch?v=VN8_LWeYlHk](https://www.youtube.com/watch?v=VN8_LWeYlHk) **Duration:** 00:17:12 ## Summary - The speaker reviewed 17 tech predictions made in January 2025, using a self‑created grading rubric to assess which were accurate (“hits”) and which missed the mark. - Seven predictions were deemed “home runs,” with the strongest being the rise of AI‑only creators who are now earning six‑figure incomes and prompting the emergence of AI‑native creative agencies. - A major shift in social media content was confirmed: roughly 40 % of Instagram feeds now contain AI‑generated material, making it increasingly hard for users to differentiate synthetic from human‑created posts. - AI integration is deepening across platforms, exemplified by tools like Google’s Genie 3 and Meta’s internal AI features, even leading to recursive AI systems where one model navigates worlds generated by another. ## Sections - [00:00:00](https://www.youtube.com/watch?v=VN8_LWeYlHk&t=0s) **2025 Prediction Review: Hits & Misses** - The speaker revisits their 17 2025 forecasts, explains their grading rubric, and highlights seven predictions that came true, notably the rise of openly AI‑generated creators. - [00:03:16](https://www.youtube.com/watch?v=VN8_LWeYlHk&t=196s) **AI Voice Companions Go Mainstream** - The discussion highlights emotional AI voice companions becoming mainstream, backlash over voice changes, rapid adoption of AI tools across non‑tech jobs and workplace platforms like Slack, and the emergence of visual AI companions driving a growing multi‑hundred‑million‑dollar market. - [00:06:22](https://www.youtube.com/watch?v=VN8_LWeYlHk&t=382s) **AI Dating Trends Self‑Assessment** - The speaker reviews their recent hits and partial successes, then predicts that AI‑powered virtual companions will expand in dating platforms despite social etiquette and ethical challenges. - [00:10:50](https://www.youtube.com/watch?v=VN8_LWeYlHk&t=650s) **Reassessing AGI Timelines and Debate** - The speaker reflects on delayed AGI expectations, the increasing oddities in AI performance, and a shift toward personal AI safety concerns. - [00:14:01](https://www.youtube.com/watch?v=VN8_LWeYlHk&t=841s) **Review of Missed AI Forecasts** - The speaker evaluates a series of optimistic AI predictions—home visual feedback tools, a robot “iPhone moment,” Super Bowl robot ads, laundry‑folding bots, and AI weddings—admitting most missed their mark, noting only companionship remains a viable prospect. - [00:17:07](https://www.youtube.com/watch?v=VN8_LWeYlHk&t=1027s) **Request for Grading Feedback** - The speaker asks viewers to identify any missed points or grading mistakes and share their thoughts in the comments. ## Full Transcript
Back in January of 2025, at the
beginning of the year, I put my money
where my mouth is or I put my
predictions out on the table. I listed
17 predictions that I thought would come
true in 2025. And just about 8 months
have come off the clock at this point.
It is up to me to come back and be
honest with you about what's working and
what's not with my predictions because I
believe in laying the cards on the
table. All right, so we're going to go
through all 17 predictions. I'm going to
share with you my grading rubric first
and why I'm grading things the way they
are for each and you can push back and
tell me where you think I'm wrong or
where you think I missed something that
might shift the grade. So, first we're
going to go through what I call the
hits, the things that I got right. So,
out of 17 predictions, seven, we'll call
them home runs. I knocked them out of
the park. They absolutely were. Number
one,
AI creators come out of the closet. I
predicted that we would see a world
where AI characters and creators compete
with human creators. Yes, I'm really a
human. I know people sometimes think I'm
an AI. I'm an actual human. The beard is
real. But AI creators are coming out of
the closet. There are AI only fans
creators that are earning $10,000 a
month or so with apparently 300% revenue
growth. So, I'm sure it's much larger
now. There are AI native creative
agencies out there now. Major platforms
are allowing AI content because frankly,
it's sticky. I'm going to give myself
yes. AI creators did come out of the
closet. This is the year of AI creators.
It's going to get bigger from here.
Maybe they're going to have an AI Nate
by this time next year. Who knows?
Number two, your feed goes synthetic.
So, this is sort of a correlary to
number one, right? If you have more AI
creators, you have more synthetic
content in your social feeds. Well, that
one is predictably a hit. Apparently, we
are nearing 40% of Instagram feeds being
heavily AI influenced. So creative
elements with AI, AI creators
themselves, etc. AI content generation
tools are being integrated deeply into
meta so Zuck can get people making more
stuff with AI. And increasingly users
cannot distinguish AI from human
content, which comes down to advances
we've seen this year in image and video
generation. If you haven't seen Genie 3
from Google, I encourage you to look at
some of the videos. It can generate
immersive worlds as you navigate them,
right? It's one of those things that's
really eerie because the world literally
doesn't exist until you go there and
then it's just it appears. And one of
the things we learned from Google in
August is that they now have so much
confidence in Genie3 that they've
actually got a tiny little LLM
navigating the world inside the mind of
that other LLM. In other words, they
have a tiny LLM steering an agent
through the world created by the AI
Gen3. There's an AI living in the mind
of another AI. I did not predict that
one, but boy would have been cool if I
did. Next hit, the revenge of audio. So,
I'm calling out sort of the importance
of her like voices and how that's going
to be a big part of 2025. We have
literally open AI's voice assistant
comparisons to her in 2025. I did not
make that one up. We have contextually
aware AI voice systems being deployed.
We have emotional AI voice companions
becoming mainstream. It is a big year
for voice. Next prediction. I don't
think that one's going to be too
controversial to say that's a hit. In
fact, it's such a hit that when Chad
GBT5 came out and the voice changed, the
character changed, the way the content
comeback changed, people complained.
They complain because they liked the
voice of the old version of 40 so much.
That's how you know it's a hip. Next,
your new colleague is an AI for non-
tech. 45% of US workers report using AI
tools at work. AI adoption is spreading
beyond tech into traditional industries.
Yeah, workplace integration is
accelerating. I'm giving myself mostly a
hit for this. I think by the end of the
year, it's going to be more true as a
hit. It's turning up that way. I will
fully agree. We are we are not in every
company at the point where you can tag
the AI in Slack. In many companies, you
can. And I think that it's worth calling
out that if that feels strange and
science fiction to you, that's more a
function of your company context than it
is of where the tech is actually at.
Next up, a visual AI unicorn or the
companionship phenomenon. This idea that
we would have digital girlfriends,
digital boyfriends, that's got to be a
hit. Grock's virtual visual companions
are rolling out on a massive scale.
People have lots of feelings and
thoughts about that, but they're there.
The AI companion market as a whole is
hitting $120 million plus per year with
widespread adoption. It's just exploded.
And so, I think that one's a hit. It's
going to be even more of a hit in four
months at the end of the year. Memory
changes everything. Chat GPT is
continuing to invest in memory. We now
have Claude investing in memory. Cross
conversation memory is happening. AI
agentic memory is happening with
multiple AI agents having different
memory strategies. We have tremendous
investment in the ability to even
partially remember things. By the end of
the year, we're going to have even
better memory features across more LLMs
and more agents. Memory is going to be a
really, really big deal. And so I'm
going to give myself a hit on that one.
I think that's one of the defining AI
transformations in 2025. And then the
great authentication crisis. Essentially
what I suggested is that we are going to
have more and more issues with these
voices creating deep fakes causing fraud
issues etc. It's certainly true that
perfect def deep fakes are becoming
trivially easy to create. It's true that
identity verification is becoming more
and more of an issue. we are having to
go to greater and greater lengths to
prove our identity. And the last piece,
the piece that I don't know if it's
actually going to come true or not. This
was a very specific prediction. I want
to be honest about the fact that I don't
know that we have this yet and we'll
have to just flip the coin for four
months to see if it really lands.
Despite the overall trend being in my
favor here, I did predict that a CEO
would have an issue because someone
would voice print that CEO from a public
earnings statement and then use that
CEO's voice to defraud a corporation.
Maybe sounds a little sci-fi. I think we
have all the building blocks on the
table in August of 2025. No one has yet
gotten that to happen that I have seen.
On the other hand, the devil in me wants
to point out if it did happen, would
they report it? I don't know. Anyway, I
think the trend is in my favor on this
one. Now, we're going to move from the
seven things that are broadly hits to
things that I want to grade myself down
on. These are partials. I think that we
have we have partial hits on all of
these. Uh there's six of them, and I
want to go through them. By the way, if
you're if you're tracking at home, seven
hits out of 17 plus six partials. Yeah,
we're not doing too bad. That's that's
13 out of 17. that that are going okay.
Let's be honest, no one's perfect,
right? All right. Uh, strong partials,
AI joins the party. So, AI, human dating
platforms launching virtual companions
and social settings. This is starting to
happen. I think that the blocker here is
actually social and etiquette, and we
will just have to see how long that
takes. I don't think the blocker is
necessarily technical, although we may
need a breakthrough app that makes it
socially acceptable for people to bring
their digital companion to lunch with
the parents or whatever. That may sound
weird and it's going to bring up all
kinds of ethics and norms issues. It's
going to happen. And so, I also want to
call out areas where like on these ones
that I'm not entirely like really seeing
this year, I want to call out like am I
doubling down here or am I walking away
and saying I was wrong. I think I'm
doubling down. I think we're going to
see more here. AI skills ultimatum. So
AI literacy is becoming a job
requirement for some roles, not for all.
Workplace de workplaces are starting to
divide over AI adoption. I see that
anecdotally with teams training programs
are proliferating. I think what makes
this a partial is that this is happening
in particular sectors of the economy.
It's not necessarily happening
economywide. And I think that's one of
the really interesting lessons of 2025
is that AI transformation is as jagged
as the intelligence of AI itself. And I
don't think I fully laid that out and
appreciated that in January of 2025. So
I'm giving myself a partial here. And
I'm going to say I am betting that the
jaggedness will continue. I am betting
that we will see these AI skills
ultimatums and AI transformations in
pockets. And I am betting that there
will be other pockets that are much less
changed over the next year or two. So I
think that I am revising my thesis a bit
here. Number three, big tech agent
fiasco. I predicted there would be a bit
of a bubble pop here. And I think that
the reason that this is a partial is
because I specifically predicted
somebody would get this so wrong there
would be billions of dollars in market
capitalization losses. And I can't prove
that there has been. Again, people are
probably incentivized not to say that
too much. So maybe there has been and I
just got over optimistic about the
economic consequences or overly
pessimistic as it were. That being said,
the larger trend, big tech did lean
really heavily on the Asian hype train
and we are seeing a lot of pullback on
AI agents because the knowhow to
successfully implement an agent that
delivers ROI is way behind the hype. The
number of people that can do a great AI
agent, even one as simple as an NADN
agent, is a tenth a tenth
of the demand at least. And so 40%ish of
AI projects are likely to be canceled in
the next uh year or two uh according to
widespread reporting um from AI newsshub
and other sources. Uh you have had very
public failures of AI agents and Gartner
itself has predicted widespread agent
shakeout. And so to me, it's the classic
Gartner hype cycle of the trough trough
of disillusionment and then we get to
the plateau of productivity. AI agents
are headed for the trough of
disillusionment. And I say that as
someone who's very bullish on agents
long term. You're just going to have to
go through the fact that a lot of people
got hyped up on it, weren't given the
tools, could not find people who could
actually build and successfully
implement them, and got very
discouraged. We need more people that
can build successful agents in
production. So that one's a partial. I
don't think that we got the billions in
market cap losses, but I think that we
did get a little bit of an agent
implosion. So, hardware, I am giving
myself a partial here because this is
behind schedule. I am doubling down on
this. I am predicting that this will
happen. I'm predicting a new specific
hardware device from OpenAI or from
others as well. Hardware with AI native
is coming. It is going to be an iPhone
challenger of some sort. perhaps not
immediately positioned to cannibalize
the iPhone, but certainly intended to be
another major device, a third device
along with your laptop and your phone.
This is coming. So, I think I'm giving
myself credit in doubling down. It is
delayed. I don't think that we are going
to see something until lateish in 2026
at this point. And so, I can't give
myself full credit here, even if I'm
bullish long term. Next one. The AGI
debate gets weird. I think that this is
partial only because of the definition
of weirdness and also there's no
consensus on AGI definition which I
probably should have emphasized more.
You do see AI systems achieving high IQ
scores and yet they also fail some basic
tests. Famously Andre Carpathy called
out at YC school that you sometimes have
AI still not realizing that 9.11 is a
smaller number than 9.89 89 because it
just doesn't look that way if you're not
reading correctly and you're not doing
the actual math. So, the AGI debate is
absolutely getting weirder and weirder
the more releases we have. And so, I
think I'm going to double down on
continued weirdness. I think that maybe
I will maybe I will back away a little
bit from the AGI debate being as central
as I think I positioned it in January. I
don't think AGI is going to matter as
much as I thought. Even if the debate is
going to continue to get weird. Okay,
last one that's a partial and then we'll
get to my misses. I know you guys stayed
around for the misses. We'll get to
that. AI safety gets personal. So,
there's a lot of like early motion
toward this. I just think I was too too
ahead of the curve. This maybe giving
myself too much credit, but like I don't
see this fully realized yet. So, there's
starting to be some AI offspaces and
detox programs emerging. We're starting
to see some public health implications
of overaffirming AI getting reported
widely. Personal AI boundaries are
becoming something people talk about.
Speaking of companions, it's becoming an
issue in human romantic relationships.
Do you have a digital companion? Do we
have boundaries around that, etc. Uh,
wellness industry is pivoting toward AI
concerns as well, but it's not really
fully there. To me, this feels like I
want to double down, but it feels like
it's running a little bit behind where I
thought it would be. And so, I think
it's more of a 2026, 2027 thing to get
widespread. Okay. Okay. I know you were
waiting for this. The misses. What has
been too early and too wide? Sort of
just not on the table. We have four
things cuz remember we're doing seven.
We did seven that were hit, six that
were partials. Now, we have to do the
fours that were terrible that I missed.
I predicted that there would be AI
consciousness activism at board meetings
from people who believe in AI
consciousness. There are certainly
people who believe in AI consciousness,
but it's been mostly an academic
discussion. It's been mostly a chat room
discussion. It has not yet disrupted any
board meetings. It seems somewhat low
probability that it will. It's I'm
giving myself a miss on this one. I
think I just did not correctly predict
this. Even though I did predict that
there would be some degree of
consciousness conversation, that's not
the same thing as saying that there
would be a sort of disruptor that's like
allin for AI labor rights at the board
meeting. I just that didn't I think the
smart mirror that I predicted was a miss
too. I predicted that there would be a
lot of investments in smart mirrors
because people want to get into visual
AI, right? They want to try on outfits.
They want to exercise things that they
can do at home that need feedback that
are visual from the AI. We've had a few
fitness and health things. It's not It's
not a major breakthrough. It's not on
track to become a household product. I
just missed this one. I'm not going to
double down. Third miss, humanoid robots
will have their iPhone moment. I
predicted specifically that there would
be a robot 2026 Super Bowl commercial,
which I still have time on the clock
for, so I'm not giving up on that. But I
think largely speaking, we still don't
have the iPhone moment for robots. We
are we are still a year or two away and
it's been that way for a bit. I know
there are now production robots on sale
that can solve household problems. I saw
one that folds your laundry. Looks way
too expensive to fold my laundry. I
think I was too early on this one. I'm
not going to double down. Last one. AI
weddings. I predicted that an AI
relationship moment would turn into a
wedding. As much as we've talked a lot
about companionship and the social
implications of AI, I've seen no
traction on AI weddings anywhere in the
world, no legal framework, nobody
stepping up to say, "Come to our
country. We will do AI weddings for
you." It might still happen, but I think
I just jumped in the water too early on
this one. So, I'm not going to double
down there. I think the companionship
thing is sticking around. I think
weddings are hard to predict. So, there
you go. Uh 41% of my of my predictions
were broadly correct. 35% eh give
yourself some partial credit and 25% 24%
no good at all just didn't get it right
what do you think about what's next what
would I what would I wish I had added to
these 17 predictions that's the last
part of this if I if I look at what has
happened so far if I could go back and
do it again what are three things that I
would I think first and foremost I
should have predicted the kind of LLM
affinity that we saw with the chat GBT5
roll out where people just demand anded
40 to come back. I had all the
ingredients there. I talked about
companionship, emotional bonding. I
should have seen it. I didn't see it.
Didn't predict it. Don't get to credit
it. Another one that I should have
predicted. I should have leaned heavier
on this being the year of code. AI and
code is something that has been so
heavily invested in by every single
major lab and frankly by a lot of the
tool builders themselves. And I just did
not put enough effort into sort of
investing in smart predictions there. I
think if I had been smarter, I would
have seen the relationship between code
and agents. I would have seen the
importance of code in flagship releases
do it again. All the ingredients are on
the table. I could have seen it. And
then three, like stepping back, I did
not correctly predict the downfall of
llama and the proliferation of open-
source. There are now hundreds and
hundreds of open source models that are
excellent. And you may hear about open
AI models, but there are lots and lots
of Chinese models that are incredible as
well that are open source, open weights
that you can then run on your own local
machines, and you don't have to run on
Chinese servers at all. I should have
done a better job noting how quickly LLM
technology proliferates and that was a
miss on my part, too. So, there you go.
Three misses, things that I think I
should have predicted, the 17 original
predictions and how I broke them down.
What do you think I missed? Where do you
think I graded wrong? Let me know in the
comments.