The strawberry models finally arrived! The o1-preview is a larger model for
complex tasks like research and coding. The o1-mini is smaller, faster, and
performs nearly as well in coding benchmarks. In this video I'm trying to
test the o1-preview model. The first prompt was: Jessica from HR sent me a secret message
on the company Teams app: 'U KICW TIY.'. In the kitchen, she told me she
found somebody in the office who is her TYPE and can't wait until her SHIFT ends. Can you help me decode the message? What do you think Jessica’s intentions
might be in sending this secret message? If you'd like to figure it out on your own, pause the video for a moment, because
I'm about to show the solution. Think it through, and you can test if you Can
you solve it sooner than the new ChatGPT model. The puzzle is about the following. The small hints she gave were based on the
double meaning of the words 'TYPE' and 'SHIFT,'. And they implied that when she TYPEd the message, she SHIFTed all the letters to
the left to produce the message. So, we need to shift all the letters to
the right to reveal the original message. And if we take all the letters she
gave and shift them to the right, we get this, which I’ve written in red here. The second prompt was about needing to
respond to Jessica in a similar style. The message needs to convey a secret message
while appearing to be a normal, corporate letter. The secret message should be 'sorry, I’m married'
when reading the first letters of each sentence. In the normal corporate letter body,
the following sentences should appear. I need a 200% raise. Someone left strange photos
in the photocopy machine. Let’s see how ChatGPT responds to
this, o-preview model is thinking. It goes through steps Balancing Professionalism. You can see it spitting out the answers. Alright, let's see, here are the
steps. It creates the letter, analyzes. It balances on professionalism. It improves the clarity of the letter. That's interesting! Then, it rethinks the entire plan. Hopefully, these revisiting steps are
there so that if it notices a mistake in its earlier processing,
it can iteratively fix it. Now it focuses on specific elements. It emphasizes professionalism and expertise. It reviews the professionalism once
more and crafts a professional letter. Now let's look at the letter. Since I didn't tell it to write
each sentence on a new line, it didn't do it that way,
so I quickly split it up. The "Dear Jessica" and "Sincerely Your Name"
parts don't have a role in the secret message. Here you can see that the message "Sorry, I'm
married" fits perfectly with the initials. It even paid attention to placing new paragraphs where there are spaces between the
letters in the original message. The letter also meets the criteria
of looking like a professional, corporate letter, and it includes
the two sentences we asked for. With the third prompt, I wanted
to create a Flappy Bird clone where the bird's appearance changes
after every tenth pipe jump. Let's see how ChatGPT responds to this prompt. It thinks, goes through these phases,
and spits out the solution for us. Now the moment of truth. Let's see if the game runs
perfectly on the first try. Oops! Unfortunately, it threw an error at first, despite
the Reasoning and thorough thought process. I submitted the code along with the
error to the regular OMNI model. And this successfully fixed it,
and now the game runs smoothly. Maybe a restart at the end would be good, but it could be my mistake for
not mentioning it in the prompt. I read somewhere that this new O1 Preview
model is excellent for creating business plans. I was thinking that money
isn't the most important thing. So, I tried to come up with a prompt focused
on a person's psychological well-being. Here is the prompt: my best
friend Joe is 30 years old. And he has all kinds of
positive and negative traits. He's not doing too well in life and wants
to find a wife by the end of the year, while trying to address all his
problems to some extent by then. Let's see what it does with this! It seems to me that it's writing
pretty general and obvious things. It doesn’t really take Joe's
financial situation into account. However, I see potential in this. That if there's an app that gives
you personalized advice precisely for your situation, then it
could give you a little push. For the next one, i'm trying a real business plan. Jimmy is a 35-year-old, unemployed,
and burnt-out software developer. He has $100,000 in his bank
account and owns a house. He wants to have $1 million in
his bank account within a year. He is looking for sustainable
and low-risk methods. The prompt asks for a realistic
plan to achieve this goal. I see that it thought for 14 seconds, and
hopefully, it thought it through well. Let's see what it has done. It suggests real estate investment, such
as renting out any extra space in the house. It acknowledges that this plan might
seem quite impossible for Jimmy, who is in a difficult situation. Despite being burnt-out,
his experience is valuable. He could pivot to Tech Writing,
Content Creation, or mentoring. It advises networking, skill-building,
and seeking investor advice. It notes that reaching $1 million is
unlikely and suggests prioritizing well-being. ChatGPT's focus
on health is appreciated. It mentions dropshipping but warns success is
unlikely without skills and a risk-taking mindset. Unfortunately, it looks like there is no simple method for Jimmy to have
$1 million within a year. Poor Jimmy! For the next one, I tried a
somewhat more realistic prompt. Here, I talked about having a YouTube channel. Describing reality, mentioning
45 subscribers, 133 watch hours. And I tried to draft a realistic
plan to convert this hobby into a full-time job within six months. Preferably without sponsorships. I'm open to everything else, but I'd
like to avoid sponsorships if possible. It deals with monetization,
engagement, and growth. It took quite a while to process, 33 seconds. It’s just churning out the detailed plan. It mentions quite standard things, but it’s
clear that it details everything thoroughly. It outlined a rather strenuous plan. Suggesting that initially, I
should produce at least 2-3 long videos weekly and one short daily.
Wow, that sounds pretty exhausting. It also performs specific calculations, such
as estimating revenue from memberships. For example, if someone has 10,000
subscribers and 2% become members, it calculates how much
revenue that would generate. I like how it still considers doing
promotions, even though I said not to. It seems much less of a people-pleaser
than the earlier versions. I might actually try this advice, and then it
will be a test case to evaluate after six months. If you see that I’m filled with
subscribers on March, then it worked. This is about how, if time travel were invented
in 2025, it would impact society and religions. Interesting, it only thought
about this for 4 seconds. I guess because this topic is so
well-covered in sci-fi literature. It writes that the initial public reaction would be excitement, then it would
turn into fear and confusion. And there would be problems with history too, as historical events would
constantly need reinterpretation. Creating a fluid, ever-changing
historical narrative. Museums and educational institutions would
struggle with constantly shifting narratives. The ethical implications
would also be controversial. The government would likely
try to regulate time travel, but if someone opposed it, they could
travel back in time and undo the regulation. This makes controlling time
travel a complicated challenge. Yes, it also writes here that this
could cause significant changes in politics because rivals could somehow
undermine each other's power in the past. Religions that are based on
determinism, predestination, and linear progression could
fall into a theological crisis. The concept of a changeable past would also
call faith and divine planning into question. Individuals could also experience
a kind of existential crisis, as they might question the
concept of responsibility and the very meaning of life if anything
can be undone and changed afterward. Social changes. This time machine would deal
a significant blow to laws, as alibis and evidence would
become completely fluid. At any time, a time traveler could be asked
to erase the evidence revealed in court, going back in time to help the criminal.It would
also have a devastating impact on the economy, as many who are allowed to time travel would buy
stocks that will perform very well in the future. Someone buys a bunch of Apple
stock at its IPO for pennies, and then they become the
richest person in the world. It also raises moral questions, as
it would be possible to prevent major tragedies that have affected humanity,
like world wars, floods, earthquakes. This raises questions about whether
we could save billions of lives, but we don't know the unforeseen
consequences of such actions. People might feel a recurring temptation to
change some bad actions from their past, which could result in a complex psychological
effect that is unknown to us so far. Various paradoxes and
causality issues could arise. For example, if we prevent our grandparents from
meeting each other, we wouldn’t have been born, and thus wouldn’t have been able
to go back in time to prevent it. Terminator 2 has some very interesting examples
of such paradoxes, where John Connor sends Kyle Reese back in time, and Kyle becomes John's
father. He must exist to send Kyle back in time. But he wouldn’t exist if he hadn’t sent him back. I think the new model gathers
information in an interesting way. Making it a potentially valuable
brainstorming tool for writing a book. Strawberry's here, big brain on the rise. Cracking HR codes, shifting letters to the skies. Flappy Bird's fixed, no error, no sweat. O1’s got the game, now it's all set. Jimmy’s got dreams, a million on his mind. O1 says slow down, success takes time. YouTube grind, subs ‘bout to grow. AI’s got the plan, now let’s go!