Pratique du Shadowing: Here's the AI Roadmap You Actually Need - Apprendre l'anglais à l'oral avec la vidéo
Choisissez un mode d'entraînement
Chaque mode ouvre une salle d'entraînement en plein écran.
452 phrases
Si les phrases sont trop courtes ou trop longues, cliquez sur Edit pour les ajuster.
1
Learning AI isn't a straight path anymore.
2
Actually, it's old school RPG themed.
3
Everybody starts off on tutorial island.
4
Those are like the basics of AI models, prompting, and using generalized AI tools.
5
But then the map opens up and you get to choose your class.
6
There is the strategist, the rogue, alchemist, crafter, sage, and machinist.
7
So by the end of this video, you will have the full map, the skill tree for your class, quests worth doing, and achievements worth unlocking.
8
Plus I will include a prompt that helps you determine your class
9
and the right quest log for you based upon your learning style and your goals.
10
Sounds good?
11
Let's go!
12
A portion of this video is sponsored by Crusoe.
13
All right, so everybody starts off on Tutorial Island in which everybody picks up a starter pack.
14
These are skills that you need regardless of what class that you choose.
15
Here is the skills tree.
16
The first skill is to understand generative AI, the basics of it.
17
What is it and what can it do?
18
Next, you need to learn how to prompt correctly because this is the way that you talk to these AIs.
19
Because if you can't communicate, how are you going to get it to do what you want?
20
Then we need to learn the general tools.
21
Tools that everybody should know how to use like chat, deep research, image video audio generation, and how to incorporate the right data.
22
This allows you to integrate AI into your day -to -day life.
23
So the good news is that there are a lot of quests, resources to choose from.
24
The combo that I first did like many moons ago was the Google Essentials combo.
25
These are two courses on Coursera, the AI Essentials course and the Prompting Essentials course.
26
These two courses cover all the skills for you to get your starter pack.
27
Very popular, sub 10 hours.
28
You can probably get this done just during a weekend.
29
The learning style is video -based with some assignments and interactive components to it, but honestly, mostly just video.
30
And it takes less than 10 hours to cover both courses.
31
And price -wise, it requires a Coursera sub for around $49 per month, but you do get a seven -day trial.
32
So I quite enjoy taking these courses.
33
I even made videos summarizing them.
34
However, more recently, Google has come up with a new certificate program called the Google AI Professional Certificate.
35
And I think if I had this option, I would take this instead.
36
It currently has eight different courses, going from AI fundamentals to brainstorming, planning, research and insights, writing, content creation, data analysis, app building and app deployment.
37
So the breadth is much wider.
38
It is primarily still video based, but I think it's more interactive because it has more activities sprinkled throughout the certificate.
39
It would take you around 13 hours to complete this.
40
And price -wise, it's still around $49 per month using Corsair Plus with a 7 -day free trial.
41
So yeah, personally, if I were to do this today, I would go with this certificate as opposed to the combo from earlier.
42
But with the caveat that this is a lot more Google -flavored.
43
Like, they focus a lot more on Google ecosystem tools, while the Google's Essential combo was pretty tool -agnostic.
44
So just something to keep in mind.
45
Another quest path that you can take
46
if you're into the video plus some exercises style of learning is the Anju Ning path, aka deeplearning .ai.
47
As many of you guys know, Andrew Ning is like a superstar AI educator amongst many other things that he does.
48
And the two courses
49
that will cover the starter pack skills would be Gen AI for everyone followed by AI prompting for everyone.
50
Take you about 12 to 15 hours to cover both courses.
51
And price -wise, it's actually free to audit all of the videos.
52
But if you want to have access to the labs, quizzes, and certificate, you need to pay for a deep learning AI pro,
53
which is around 25 to 30 dollars per month.
54
I would say pick this path if you want something that's more tool agnostic
55
and more not theoretical but more about like frameworks
56
and like big pictures because Andrew is just so knowledgeable and experienced in this field.
57
He's able to give you insights about the industry
58
and the specific skills in a way that I would say is pretty unmatched
59
but he does have a pretty specific way of teaching.
60
So if you're interested in this path I would definitely recommend checking out some of his videos
61
and teaching style first before committing.
62
By the way just to be super clear I'm presenting a lot of different options for you
63
so you can make your best choice based upon your budget,
64
your time, and your learning style but you don't need to be doing like multiple different paths.
65
You just need to choose one and you may get all the foundational skills that you need.
66
Okay so the next choice you have is called the Elements of AI by the University of Helsinki.
67
The format of this course is interactive text with exercises.
68
Also very popular.
69
It takes around 30 to 50 hours to complete so it is a big course.
70
It's also free and includes certificate as well.
71
Great option if you're looking for something very thorough, free, and you're more into the interactive text -based learning style.
72
Quest path number five is the Anthropic Claude Path.
73
Anthropic has a series of courses that covers all the skills needed for the starter pack, including CLOD 101, AI Influency,
74
Framework, and Foundations, AI Capabilities and Limitations, as well as Prompting Docs.
75
By the way, I'll be linking all the course links in the free resource in the description.
76
So no need to like try to write down all the course names.
77
Anyways, yes.
78
So the format is interactive, structured lessons with quizzes.
79
Probably take you around 10 to 15 hours.
80
It's also completely free and you get a badge per course as well.
81
However, the caveat here is that because is produced by Anthropic.
82
Naturally, the course heavily flavors Clod and Anthropic products.
83
So having done these courses, they do teach you like really great foundations as well,
84
but obviously they'll be teaching you based upon like the Clod stack.
85
And of course, we also have to talk about Clod's biggest competitor, which is OpenAI.
86
OpenAI also has some really great resources, specifically their AI Foundations course and the Applied AI Foundations course.
87
These two courses will cover all the skills you need for the starter pack.
88
It's also a self -paced exercise and assessment based content is great obviously also has a very strong bias towards OpenAI products, which makes sense.
89
So I would recommend this path only to people who are very into the OpenAI ecosystem.
90
It is also completely free.
91
You also get a badge per course.
92
And if you do those two courses, plus there's a third one called Agents and Workflows, you would also get a certificate.
93
Time will also take you around two to five hours to complete.
94
So shorter than Deanthropic courses.
95
I also got to say that the vibe of Deanthropic courses with the OpenAI courses are really, really different.
96
Now, of course, there are a lot of other courses and options.
97
I just try to pick what I think is the most comprehensive,
98
but also give you guys like different options based upon learning style and budget and things like that, right?
99
So I'm going to put on screen some of the other options that you can check out as well.
100
And there are two other options I want to do a quick shout out.
101
The first one is the Kaggle five -day
102
Gen AI intensive with Google.
103
This was a live and recorded cohort with different project assignments and very community based.
104
I really love this because it's like a five -day intensive situation in which you cover all the basics.
105
And there's a capstone project in the end too.
106
It's also completely free.
107
You get a Kaggle batch too.
108
However, they only did this one time.
109
It seems like they're probably not going to do it live again, but you can't access all of the recordings and go through it.
110
So if you're someone who prefers cohort -based intensive live sessions, I would recommend checking this out.
111
Although it's no longer live and maintained, sadly.
112
The other path that I want to give a shout out is the Careers Essential in Genitive AI by Microsoft and LinkedIn.
113
This is a very, very popular course back in 2024.
114
It's video -based plus a final exam and Microsoft -flavored.
115
Takes around six hours to complete and also free.
116
However, it is quite outdated.
117
I went on their landing page and it says that they're going to be updating it in September 25th, 2026.
118
I filmed me this video prior to that so I haven't been able to check out this new updated version yet.
119
But by the time that you see this video it will definitely be after 25th
120
so if they actually release this course I would recommend checking it out especially if you like
121
Microsoft flavor things.
122
All right here is the summary slide again with all of the different options
123
that I talked about and ones that I didn't explicitly mention too.
124
I will also include a prompt in the description
125
which you can copy paste into your favorite AI chatbot to help you figure out
126
which quest path is the best for you as well.
127
Hopefully that's helpful too because there's just like so many resource is an option.
128
It's good that there are so many options, but you know, paradox of choice.
129
All right, amazing.
130
So at this point, if you complete one of these quest paths, you would have gotten your starter kit and you're ready to leave tutorial island.
131
Now you get to make the very exciting choice of which AI class you want to belong to and level up in.
132
There is the strategist, rogue, alchemist, crafter, sage, and machinist.
133
I will go through each of these classes, their skills tree, and the quest paths, the resources that you can choose from.
134
the first class I will cover is the strategist the strategists are the people who are interested in AI leadership
135
and strategy your goal is to understand AI in a way
136
that leads adoption and sets the AI direction in your organization you are the AI thought leader
137
and your skill tree after getting the starter pack of course
138
starts off with understanding the vendor landscape like Google Microsoft AWS OpenAI Anthropic
139
so many different options what do they actually offer and What is the most suitable for you, your business, and your clients?
140
You also need to acquire the skill of how to run a team with AI.
141
And this is actually very non -trivial.
142
You can't just give your colleagues, your employees, the people you work with AI tools and just be like, hey, let you guys figure it out.
143
That doesn't work.
144
You need to know how to create human AI agent teams.
145
What are the delegation norms?
146
How do you evaluate them?
147
You also need to learn the skill to set strategy.
148
Where is it that AI actually matters?
149
What's the AI roadmap for your team?
150
And of course, you need to lead this adoption, the execution part.
151
How do you get buy -in from everybody you work with?
152
Your investors and how do you actually get your team to adopt your AI roadmap and strategy.
153
Doing this while also balancing, of course, business metrics like ROI.
154
Now let's talk about some quests, some paths to choose from.
155
Now, if you're someone who isn't already like deeply affiliated with a specific vendor
156
and someone who is not necessarily really into certificates, like official certificates either, I would actually recommend the AI Leadership course by OpenAI.
157
It is a self -paced, text -based, interactive course with different activities and an assessment.
158
And I think it actually really does a good job of covering what AI leadership looks like
159
and different considerations and frameworks that you can apply, including business priorities, governance, stakeholders, how to come up with an adoption roadmap,
160
and how to draft an initial AI strategy.
161
And it does this in a pretty objective way, regardless of the type of business that you have and the type of tools that you're using.
162
It'll take you around three to four hours to complete, and it is free.
163
And after the AI leadership course, you can also supplement it with a very, very popular course called AI for Everyone from Anjone.
164
I believe this is literally like the most enrolled AI course ever.
165
Takes around seven hours and it's free to audit.
166
But the catch on why I say this is optional is
167
because this course was actually created prior to the Gen AI boom.
168
So a lot of things are pretty outdated, although a lot of the organization and team building frameworks that Andrew gives is evergreen.
169
Great, so that's the path I would recommend if you're not already like deeply affiliated with a specific vendor.
170
But if you are, no worries, because most of the major AI vendors do have their own version of an AI leadership coursework certificate.
171
Anthropik does, Google does, Microsoft, AWS,
172
OpenAI, etc. So my suggestion is to just go learn from that specific vendor and get an official certificate if you wish.
173
I'm going to put on screen now some of the different options by vendor.
174
There are also a bunch of university certificates
175
and degrees for AI leadership as well and I'll list some of the cohort -based ones too.
176
Finally, books.
177
There are two books that I would recommend that you check out if you're into reading books.
178
The first one is called Co -Intelligence by Ethan Malik.
179
Gives a lot of frameworks for organizational thinking.
180
And AI Snake Oil by Arvin Narayanan and Sayesh Kapoor.
181
This is like from the skeptical perspective about what AI can't do in organizations.
182
So it's pretty refreshing.
183
All right, I'm gonna put on screen now a summary slide now for the strategist class.
184
And with that, let's move on to the rogue class.
185
The rogue class is all about using AI agents really well and hacking things together.
186
Getting your agents to do your work for you.
187
If you're someone who's all about hacking and tinkering and speed and learning by doing, then you are probably in the road class.
188
Like creating your own custom Hermes setups, co -work setups.
189
By the way, co -work is just called Claude now, I think.
190
And your own agents and workflows.
191
And here is your skill tree.
192
First skill is knowing what agents can do.
193
What's the difference between the agent versus a chatbot?
194
When do you trust them?
195
And when do they fail?
196
The next skill is delegating your real work to AI agents.
197
Testing it out, pushing the boundaries, see what they can do.
198
Over days and weeks, you also have to learn how to make them repeatable.
199
The amazing agentic workflows that you make, you want to make sure that you can actually use them on a day -to -day basis.
200
And finally, building your own AI agent setups.
201
Learning things like a Hermes setup that runs your entire life and your work.
202
A core setup that helps you with content.
203
And all sorts of different custom agents and workflows that help you throughout your life and work.
204
Along the way, you will learn things about gateways, memory, skills, security, and how to self -host different models too.
205
Now, since the road class is all about hacking things together and figuring out your own custom agent builds,
206
there isn't really like a general course that will teach you all of these skills.
207
You learn them as you hack along.
208
So the first task of the Rogue class is to figure out
209
which system or tool or ecosystem you want to start building with.
210
It's all about mastering how to use AI agents
211
and chaining them together in a way so that you can do what you want it to do.
212
And there are a lot of different AI agents to play with.
213
You have Cloud Cowork agents, now called just
214
Cloud from Anthropic.
215
You can use agents and build multi -agent workflows with OpenAI, Microsoft and Copilot flavored, Hermes and OpenClaw
216
if you don't want to be vendor locked and if you're into using open source AI.
217
So the first thing to do is just choose
218
which kind of ecosystem that you're most attracted to
219
and go through those specific documentations and tutorials so you learn how to use those agents and build with them.
220
I'm going to put on screen now some of the different options and what each ecosystem slash tool specializes in.
221
And I want to mention there is one cohort -based course
222
that seems to have pretty good reviews that teaches you how to use agents as well.
223
I have not taken this course.
224
I have not been part of this cohort.
225
Never used Omarki either.
226
Apparently, it's like an ecosystem tool set, but just wanted to put it out there for anybody who wants cohort -based stuff.
227
There really isn't any certificates or books in the rogue class just because of the nature of it.
228
It's so like hacky and custom.
229
So if you're in the rogue, alchemist, sage, or machinist class, customizing and building with open source AI as well is a must in my opinion.
230
But it can be such a pain dealing with the infra surrounding these open models though, which is why when the time comes,
231
I highly recommend that you check out Crusoe Intelligence Foundry.
232
It does two things.
233
Serverless inference allows you to run the leading open models without managing any of the infrastructure.
234
And serverless fine -tuning lets you customize these open models with your own data without having to own any hardware.
235
Let me show you.
236
So here's a data set for my notes from the AI courses that I've been taking.
237
I upload it, pick an open model, and start a fine -tune. And voila!
238
Just like that, I have this custom model that's trained on how I study personally
239
that I can now use for all sorts of things.
240
Like building a personal study app, for example.
241
Customizing a model like this used to mean provisioning clusters and finding setups for hours.
242
But now you can have a working model in just minutes, which you can call from any device.
243
Even as small as an Arduino or a Raspberry Pi.
244
And build with them too, if you like.
245
There is so much that you can do.
246
So do sign up now for Crusoe Intelligence Foundry.
247
Because they're currently also giving away $5 in free credits.
248
Link is in the description.
249
Thank you so much, Crusoe, for sponsoring this portion of the video.
250
Now back to the video.
251
Alright, let's move on to the alchemist class.
252
Now the rule class is all about using AI agents
253
and hacking them together and chaining them and customizing it in the way that you want.
254
The Alchemist class is about building AI agents and turning them into sustainable and reliable systems, usually as part of other business workflows.
255
And this is the skills tree.
256
First, you're going to learn the fundamentals of agents since we're going to be building them.
257
Then you want to build your first agent.
258
Learn to give it tools and knowledge.
259
Orchestrate multi -agents.
260
Learn about evals and reliability, which is like the other half of building AI agents,
261
is that you need to be able to assess them and how to put them into production.
262
How to deploy them, monitor them, calculate costs, and make sure that they play well with the other workflows that you may have in your business.
263
If you wish to be and to progress in the Alchemist class,
264
the first decision point is if you want to go the no -code or the code route.
265
The good news is that there are a lot of tools and resources and paths that support both of these routes.
266
Let me cover the no -code route first.
267
Step one is choosing a foundational course that covers how AI agents work.
268
The two courses that stand out the most in this category is the Google Skills Agent Fundamentals course
269
and the Vanderbilt Agentic AI and AI Agents course.
270
Both have really great ratings.
271
I will put their stats on screen.
272
Now, if you choose to go to the code route, if you're someone in the alchemist class, you will be building your own agents.
273
At Lonely Octopus, we run an AI agent bootcamp ourselves.
274
And during and after the program, we see people build all sorts of AI agents that support specific workflows and businesses.
275
Like for example, if you're a lawyer, you can be building case -specific AI agents that help you analyze your cases.
276
If you're someone in the medical field, we've seen people build agents that support triaging, handling medical paperwork.
277
Apparently there's a lot of medical paperwork as well as continued education.
278
Now to be able to build these workflows,
279
not only do we need to have to understand how AI agents work and know how to build them,
280
you also need to learn how to integrate them into existing workflows.
281
Now if you choose to go the code route, you have a lot of different choices and resources.
282
So many different quests to pick from.
283
There is the Deep Learning AI, Agentec AI course by Andrew Ning, which is video and assignment based.
284
There's Hugging Face, which really emphasizes on the open source side.
285
And it's a combination of text and code, very interactive.
286
There's also the Google and Kaggle cohort based five -day AI agents intensive, the Vibe Coding Edition.
287
And yes, so many more.
288
Next, regardless if you chose the code or the no code option,
289
you have the optional quest to go deeper onto a specific platform that you want to be shipping your AI agents on.
290
That can be Anthropic, OpenAI, Google, Microsoft, AWS, Langchain, and so many other open stores and third -party platforms too.
291
So basically you just want to pick which ecosystem or platform
292
that you want to be using and then go through their resources and start implementing everything you learn from the foundational course.
293
And finally step three, required for everybody as well, is to learn about reliability.
294
Reliability is something that is so underrated and the reason why
295
so many companies struggle with actually incorporating AI into their existing workflows.
296
So I want to have a special emphasis on this.
297
There is Lang Chang Academy that goes through this, a cohort -based Maven course, and the book called AI Engineering by Chip Wen,
298
I think, is fantastic as well.
299
Great.
300
I should now cover all the skills that you need in the Alchemist class.
301
However, people who are of the Alchemist class are people who generally are working as either contractors or in existing businesses.
302
So certifications can matter.
303
So I'm also going to put on screen some certificates
304
and also exams that you may be interested in
305
if you want to get a certificate that proves that you have mastered the alchemist class.
306
Now here is the summary slide for the class including some honorable mentions too.
307
Amazing!
308
Next up crafter class.
309
Crafters are builders, specifically builders of AI native apps and products.
310
If you can master the skills of a crafter, you can build all sorts of different apps,
311
consumer products, b2b products, all at lightning speed with the help of AI tools and coding agents.
312
If you are a crafter, you are a builder.
313
And here is the skills tree.
314
First of all, got to know the foundations of AI coding.
315
There's a lot of names for this like vibe coding, AI assisted engineering, agentic coding.
316
Basically, there's a skill about how to use AI in order to code better and faster.
317
And the interesting thing is, is that you can be a crafter without actually knowing how to code.
318
Interesting.
319
I'll talk a little bit more about that later.
320
But first, let's go through the rest of the skills.
321
Then you need to pick your tool, whether that be codex, cloud code, cursor, co -pilot, lovable, bolt.
322
Crafters need their tools and then they need to master their tools.
323
Learning how to use that tool to direct a coding agent in a development process.
324
And as you keep building, you probably also want to start incorporating AI into the products that you're building itself, kind of meta.
325
And that's when you also need to learn how to build with AI, how to build using AI and also how to build with AI.
326
And finally, you need to learn the skill of how to ship responsibly, how to deploy the things that you build, make sure that they're secure.
327
And how do you work with a team of crafters?
328
Now, unfortunately, actually quite surprisingly, there really isn't any well -established crafter AI coding foundational courses.
329
Like there are some, but I don't think any of them are very, very thorough.
330
And maybe it's not so surprising because the field of AI coding just moves so quickly.
331
So that's what we do need to mostly rely on tool specific, vendor specific resources for learning how to be a better crafter.
332
And the first decision point that you would make in the crafter class is to decide
333
if you want to go no code route or decode route to pick the tool that you want to use.
334
I will put their stats on screen.
335
Next, regardless if you're going to code route or to no code route, everybody then needs to learn about guidelines.
336
How to sustainably build, deploy, and monitor what you build.
337
For that, there is a book called Beyond Vibe Coding from Coder to AI Era Developer, which I think is a pretty good book,
338
but it is technical and assumes that you know how to code.
339
So if you're a no code crafter, that's okay.
340
You just need to skip some of those parts.
341
And finally if you're someone who works in the team of coding crafters
342
or someone who manages a team of crafters i would recommend
343
that you check out anthropics ai native sdlk
344
which stands for software development cycle playbook it is anthropic
345
and clod flavored
346
but it has some really great principles for how to run teams to ship as fast as anthropic does
347
because they ship really really fast okay
348
i'm gonna put a summary slide on screen now we'll also have more details
349
and links and the prompt to help you figure out the tool that most suits you in the free video companion below
350
And with that, we'll move on to the SAGE class.
351
Those of the SAGE class are the ones that really understand how these AI models work.
352
In fact, they are the ones that build these models as well as customize them, called fine -tuning.
353
They are deeply knowledgeable and the backbone of what everybody is building on top of.
354
And here is the skills tree for the SAGE class.
355
First, I got a quieter skill of how AI models actually work.
356
And we're not just talking about fundamentals here.
357
We're talking about the technicalities too.
358
How models actually work.
359
What even is gradient descent and back propagation?
360
You also need to learn how to train neural networks.
361
Specifically, you become intimately familiar with the transformer, which is what an LLM actually is.
362
A LLM is a transformer that is trained on text data at scale.
363
After this, there's also the skills of fine tuning.
364
How do you customize models?
365
Inference and deployment.
366
How do you make the models usable by other people?
367
And if you are the sagiest of sages, you would also be pre -training from scratch, actually creating your own models.
368
You better have a lot of money to do that though.
369
So what are the quests?
370
Well, first off, we need to start off with the foundations.
371
You need to learn how the models work, the underlying technology, and how to implement them.
372
Luckily, there are a lot of different resources and paths that you can take to do this.
373
There is the Andre Kaparthi Neuro Network Zero to Hero course, where you literally build neural networks from scratch using code.
374
It's also a Stanford course, an animation series from 3Blue1Brown.
375
So gotta choose one of these first.
376
Then step two is to pick a track, a decision point.
377
Do you want to focus on fine -tuning, which is customizing models that are already existing, or building from scratch, literally making your own AI models?
378
I will put on screen some of the resource options that you have for both of these tracks.
379
Then regardless of which track you're on, you need to learn the skills of deployment, how to make your models usable by other people.
380
Interestingly, there's actually not that many structured courses for this because it is a pretty niche skill.
381
A lot of it is just like demos and documentation and just figuring it out.
382
But I will link some resources to check out in the free guide below.
383
And that will cover most of the skills that you need.
384
However, if you are someone who cares about certification, which you probably should care about that,
385
if you are in a SAGE class and not just like doing this for fun for yourself,
386
I will put on screen some certifications and exams primarily from NVIDIA
387
that certify you that you are a true and proper sage.
388
I will put on screen now a summary of the sage class, including the resources and some honorable mentions as well.
389
Then that leads us to the final class, a class that is shrouded by mystery actually, and that is the machinist class.
390
The machinist class is in the intersection of hardware and software.
391
They care about physical and local AI.
392
The nitty -gritty how to run and scale AI on physical devices.
393
I have personally been exploring this for the past few months.
394
Learning how to build with hardware, learning how to run models on hardware, and how to use AI models to control hardware.
395
It has been a really fun experience.
396
I have a little video here that shows some of the things I've been doing.
397
But if you're someone who's interested in local AI, running on different devices, having it interact with the world,
398
making AI -powered robots, just combining together AI with hardware, then you may be interested in becoming a machinist.
399
And here is the skill tree.
400
First, you gotta learn about the basics of how AI hardware works.
401
Then you actually want to start building with it by using either a specialized kit or different devices like the ESP32,
402
the Raspberry Pi, which is what I use.
403
And as you tinker more with that, you learn about memory and bandwidth and compute, quantization.
404
What are the different ways to run AI models on the specific type of hardware
405
that you have and how to chain together different hardware to run the models more efficiently.
406
And finally, if you really want to become the top of the machinist class, you can learn about how to scale up,
407
like how to actually run the massive AI models that companies like OpenAI and Anthropic are running?
408
What's the underlying hardware of that?
409
Like when people are talking about AI data centers,
410
you would be learning the mechanisms behind it and how they actually work and how to operate them.
411
Or you can go full on into robotics.
412
There are certain AI models and platforms that are designed specifically for physical devices.
413
And that in itself is a whole very interesting field.
414
So what are some of the resources?
415
What are some of the quest paths?
416
Well, first off, step one is to learn the basics and to start tinkering.
417
There are a few different options, but the two that kind of stand out the most to me is the first one, the machine learning systems.
418
This is two volumes of textbooks, which is completely free, including interactive labs, software, and you can buy their hardware kits too.
419
Take you probably around like 20 to 30 hours to get through all of it.
420
It is very extensive.
421
I'm actually working through it myself right now and I'm learning so much.
422
There is also an MIT course called TinyML and Efficient AI Computing.
423
This is the university track.
424
This is for you if you are someone who likes to learn in a university way.
425
It covers all the theory really, really well and includes different assignments.
426
It's about 50 hours to complete.
427
Then after this, we become more practical.
428
How do you actually get these large language models and different types of AI models to run on these devices?
429
The major player here is NVIDIA.
430
Not only can you learn this knowledge and apply them, they also do offer certificates if that is something that you care about,
431
which you probably should care about, especially if you're someone who's interested in scaling up or learning more about AI data centers,
432
how to operate them and all the underlying mechanisms or
433
if you're someone who wants to go deeper into robotics
434
and there is the Hugging Face Robotics course which you can learn
435
if you do not know anything about robotics but really want to go deep into it.
436
It is completely free and it's great if you want to crash course on robotics.
437
NVIDIA also has some great resources on robotics and physical AI.
438
Overall, I would say the Machinist class is a class that has the least amount of resources,
439
just because it's so new and is the intersection of both hardware and AI.
440
So do expect to be reading through a lot of documentation and just experimenting a lot.
441
That is what I am learning as I embark on this journey as well.
442
I'm gonna put a summary slide now of the Machina's class, all the different resources and some honorable mentions too.
443
And with that, we are at the end of this video.
444
Congratulations!
445
Let me know in the comments, what is your class?
446
And how far along are you in your learning journey?
447
For me, my major class is the alchemist class with a minor class in crafter and rogue.
448
And the machinist class is also one that I'm very, very interested in.
449
So yes, there's nothing stopping you from belonging to different classes too.
450
There's so much to learn and honestly, I'm having so much fun doing it.
451
And with that, thank you so much for watching until the end of this video.
452
I will see you guys in the next video for live stream.
Télécharger l'application
Tout pour parler couramment
Prononciation IAChaque phrase notée
Phonétique (IPA)Maîtrisez chaque son
VocabulaireEnrichissez votre lexique
Jeu de vocabulaireApprenez en jouant
Même chaîne
C12
Learn 10X Faster In 10 Minutes
Daily ConversationUniversity StudentsWorking Adults
B21
Updated Essential AI Skills For 2026
Daily ConversationUniversity StudentsWorking Adults
C1My Full Claude Cowork Setup (steal my workflows!)
Business EnglishWorking Adults
C11
Essential AI Skills For 2026
University StudentsWorking Adults
C12
How To Learn To Code In 2026
Daily ConversationUniversity StudentsWorking Adults
C1Every Size Local AI In 24 Minutes
Business EnglishUniversity StudentsWorking Adults
C1Best Upcoming Jobs (2026 - 2030)
Business EnglishInterview EnglishUniversity Students
FOR YOU
Vidéo recommandée
C11
197 phrases
How I use AI to learn engineering (anti brain rot studying)
Daily ConversationUniversity Students
C11
105 phrases
AI Governance Explained: Ethical, Transparent & Accountable AI for the Future
Daily ConversationUniversity StudentsWorking Adults
C11
74 phrases
LangChain in 5 Minutes (Explained Clearly)
Daily ConversationUniversity StudentsWorking Adults
C12
186 phrases
Learn Java in 14 Minutes (seriously)
Daily ConversationUniversity Students
C11
747 phrases
MIT Introduction to Deep Learning | 6.S191
Daily ConversationUniversity Students
C11
111 phrases
Spec-Driven Development: AI Assisted Coding Explained
Daily ConversationUniversity StudentsWorking Adults
C1204 phrases
Prompt Engineering - Guidelines
Daily ConversationUniversity StudentsWorking Adults
C11
345 phrases
Andrew Ng on Coding Agents and Deciding What to Build
Daily ConversationUniversity Students
C11
63 phrases
What Is AI? | Artificial Intelligence | What is Artificial Intelligence? | AI In 5 Mins |Simplilearn
Daily ConversationUniversity Students
B23
443 phrases
Complete Roadmap for Backend Software Engineers (START HERE!)
Daily ConversationUniversity StudentsWorking Adults
Vocabulaire et conseils d’expression pour cette leçon
Cette leçon d’expression orale s’appuie sur la vidéo « Here's the AI Roadmap You Actually Need ». Les mots qui reviennent le plus souvent : agent, skill, model, learn. Cette vidéo contient 452 phrases et 5877 mots à répéter en shadowing. La partie parlée dure 25:31. Le locuteur parle vite, environ 230 mots par minute : attendez-vous à des liaisons et à des sons réduits. 82 % des mots font partie des 3 000 mots les plus courants en anglais ; le reste mérite d’être vu avant de commencer.
Vocabulaire clé de cette vidéo
15 mots de la vidéo à retenir, avec leur prononciation et leur sens :
| Mot | Prononciation | Sens |
|---|---|---|
| certificate nom | /səˈtɪf.ɪ.kɪt/ | document, certificat |
| skill nom | /skɪl/ | habileté, compétence |
| quest nom | /kwɛst/ | quête |
| hardware nom | /ˈhɑɹdˌwɛɹ/ | équipement |
| sage nom | /seɪd͡ʒ/ | sauge |
| interactive adjectif | /ˌɪn.tɚˈæk.tɪv/ | interactif |
| rogue nom | /ˈɹoʊɡ/ | canaille, fripouille |
| prompt verbe | /pɹɑmpt/ | inciter |
| vendor nom | /ˈvɛn.dɚ/ | vendeur, vendeuse |
| ecosystem nom | /ˈikoʊˌsɪstəm/ | écosystème |
| summary nom | /ˈsʌm.ə.ɹi/ | résumé, sommaire |
| robotics nom | /ɹoʊˈbɑ.tɪks/ | robotique |
| regardless adverbe | /ɹɪˈɡɑɹd.lɪs/ | malgré tout, malgré cela |
| custom nom | /ˈkʌstəm/ | coutume |
| slide nom | /ˈslaɪ̯d/ | toboggan, glissoire |
Les verbes à particule que vous entendrez
| Mot | Prononciation | Sens |
|---|---|---|
| check out verbe | mater, regarder | |
| go through verbe | traverser | |
| figure out verbe | cerner | |
| come up with verbe | inventer, imaginer | |
| get through verbe | /ɡɛt ˈθɹu/ | franchir |
| open up verbe | ouvrir | |
| pick up verbe | soulever | |
| sign up verbe | inscrire |
Des phrases à répéter
Des phrases courtes et complètes de la vidéo, à réutiliser dans la conversation de tous les jours :
- How do you evaluate them?
- How do you get buy -in from everybody you work with?
- What's the difference between the agent versus a chatbot?
- When do you trust them?
- How do you customize models?
La grammaire de cette vidéo
Les structures que le locuteur utilise le plus, avec les mots exacts de la vidéo :
| Structure | Dans la vidéo |
|---|---|
| Phrases conditionnelles if + proposition, will/would + verbe — une condition et son résultat | if I had this option, I would take |
| Present perfect continu have/has been + -ing — une action commencée avant et toujours en cours | I've been taking · have personally been exploring · I've been doing |
| Propositions relatives who / which + proposition — une précision sur une personne ou une chose | competitor, which is · products, which makes · people who are |
| Voix passive be + participe passé — l’accent est mis sur ce qui arrive, pas sur qui le fait | is sponsored · have is called · is produced |
Prononciation à surveiller
Le locuteur utilise 75 contractions et formes réduites, comme you're, I'm, isn't. Prononcez-les sous leur forme courte, telles que vous les entendez.
- Les sons « th »: thorough /ˈθʌɹoʊ/, anthropic /ænˈθɹɒp.ɪk/
- Les sons « sh » et « zh »: adoption /əˈdɑp.ʃən/, shout /ʃaʊt/, certification /ˌsɜː.tɪ.fɪˈkeɪ.ʃən/, documentation /ˌdɑkjəmɛnˈteɪʃən/, optional /ˈɑp.ʃə.nəl/
- Mots longs — placez bien l’accent: certificate /səˈtɪf.ɪ.kɪt/, interactive /ˌɪn.tɚˈæk.tɪv/, ecosystem /ˈikoʊˌsɪstəm/, fundamental /ˌfʌn.dəˈmɛn.təl/, primarily /ˈpɹaɪ.mə.ɹə.li/
Les sons difficiles pour les francophones :
- /h/ — il se souffle, il n’est pas muet: hardware /ˈhɑɹdˌwɛɹ/, helpful /ˈhɛlp.fl̩/, comprehensive /ˌkɑm.pɹəˈhɛn.sɪv/, cohort /ˈkoʊ̯ˌhɔɹt/
- /tʃ/ et /dʒ/ — à ne pas adoucir en « ch » et « j »: sage /seɪd͡ʒ/, scratch /skɹæt͡ʃ/, badge /bæd͡ʒ/, knowledgeable /ˈnɑ.lɪ.d͡ʒə.bəl/, naturally /ˈnæt͡ʃ.(ə.)ɹə.li/
- /r/ anglais — langue recourbée, sans frotter la gorge: hardware /ˈhɑɹdˌwɛɹ/, rogue /ˈɹoʊɡ/, prompt /pɹɑmpt/, starter /ˈstɑɹ.tɚ/, summary /ˈsʌm.ə.ɹi/
Comment s’entraîner avec cette vidéo
- Écoutez la vidéo en entier une fois sans parler et notez les mots que vous ne connaissez pas.
- Commencez à la vitesse 0,75×, répétez phrase par phrase, puis revenez à la vitesse normale quand cela devient facile.
- Enregistrez-vous et comparez avec l’original, en faisant attention à des mots comme certificate, skill, quest.
Qu'est-ce que la technique du Shadowing ?
Le Shadowing est une technique d'apprentissage des langues fondée sur la science, développée à l'origine pour la formation des interprètes professionnels. Le principe est simple mais puissant : vous écoutez de l'anglais natif et le répétez immédiatement à voix haute — comme une ombre suivant le locuteur avec un décalage de 1 à 2 secondes. Les recherches montrent une amélioration significative de la précision de la prononciation, de l'intonation, du rythme, des liaisons, de la compréhension orale et de la fluidité.