AI 101 with Brandon Leshchinskiy

MIT RES.6-013 Neuroscience and Behavior, Fall 2021
Instructor: Brandon Leshchinskiy

This is a video of Brandon Leshchinskiy's presentation "AI 101," which provides an introductory overview of the application of artificial intelligence in solving real-world problems: which tasks machine learning handles well, how it works, and what its shortcomings are.

License: Creative Commons BY-NC-SA

Speakers: Brandon Leshchinskiy

[parikshitkumar]

I really enjoyed the lecture. One of those rare classes I didn't nod off. ❤

[pranaypallavtripathi]

Great lecture. Provides much needed overview for beginners who are attracted to AI

[anonymous.youtuber]

So much is to be learned from this video ! Outstanding !

[willminerva]

an AI is gonna stumble on this video and become self-aware

[GingerCampbell]

Despite the sound quality this is a great introduction to AI: both its promise and its limitations.

[dlrowle]

love this channel, you make the simple numbers more simplex. Q: What do you call a negative one sitting all alone in an empty room? (overnumerousness)

[DMS]

Great lecture. When he brought up the black cat and the system’s failure to identify it correctly, can it be told that no, this is a cat - and the model updates itself? Or is the only solution to completely retrain the thing?

Basically can the model continue to evolve and get better just in regular use.

[卢军宏台长心灵法-ud]

In the process of growing,
the main concern is how to overcome
egoism and dig up the roots of evil.

[veritas]

Can be structured better; you start off nicely and then assume viewers know neural networks ? :) All the best - mit alum

[Ben-oniDankyi]

How do you know the data you are using is already cleaned or have to be cleaned

[antoniobento]

39:30 "If you have a problem with drug overdoses or with suicides AI is probably not gonna help"- I strongly disagree. Can someone tell me why he says that AI can't solve human problems? I mean I there is already proof that proves AI can help on those types of problems. I really don't know why he would say that.

[beegyoshi]

Hubris to assume no general Ai exist, Tell that to my currently limited 350 tb consciousness.

[JeffersonRah]

I believe there is a spiritual and emotional part of life that machines will and should never learn. I personally think the convenience of AI is the disability that human beings will grow with.

[TheThinkinGuy]

cat is not mean looking... grumpy cat!

[gilgamarsh]

32:29 Adding that there is a white male bias... could have picked on anyone or anything else but went with white men. So tired of it being acceptable to pick on white men. It isn't and this needs to stop, Brandon Leshchinskiy.

[MasterCivilEngineering]

Fresh Civil Engineers, get hired by top construction companies of the World by mastering the concepts of practical civil engineering and take your career to the next level💯🇺🇲🇺🇲

[topnoodlepaul]

PROBLEM IS - we are clouding person's colour. Then define to not recognise a colour. Define it is a pattern in some emotes.

But now we have a cultural difference! And the emotes of some similar pattern of 'people' without skin colour are different, really different. well we clouding cultural patterns. It acts strange? have no colour, it is human... lets use moves reflexes, not colour and emotions! Cool, now we have to define how to realise if the homeless person with problems in health, undefined skin colour and cultural emotes should act like. Aggressive or friendly?

Ok lets pattern wha the voice it produces! Damn language is ideal. But the object uses in common an aggressive words in hello! Okay. Changing pattern, it is definitely moral to cloud, colour, culture, healthiness of moving and local dialect and mood of language it uses. Now how the hell you define it is aggressive or not? The person having a gun in pants and u have no statistics about his race, culture, health, mood patterns in language, indefinite emotes. Use a statistics by the whole country! Okay, it is a homeless in a safe rich HighHill.

GOOD. Let be say model knows whenever alien object crossing lines of his district of habitation which is patterned to him by the number of criminal checks. And.. we have no checks there cus its too criminal even for cops to get in to do their job. WHAT NOW? Maybe we can uncloud a pattern of skin colour and it would be easier then to predict next steps into how the persons pants usually fit on object with a gun in it and a pattern of the object stinks with piss and alcohol the model cannot recognise? Or a real world only rumors about those person has lost his last cash and crawled uphill in need to robb some white for buying lifesaving yet drugs?

Damn you. Rasist is when u are DELETING alien race. Not when you use patterns. When statistically black are 1/3 imprisoned then it is SO. Don't cloud it. Cus you aren't when u see that object when exit your fancy modern and intellectual sex-race-drug-tolerant cafe party in a rich HighHill. You momentary trash your idiot pattern clouding. Cus your cookie is your life and there is no restart. Blame me but you are racist when you won't die and go away quickly calling 911 but you havin a chance to survive even if the object gonna tell to his free state lawyer after arrest that your decision was racist. And you're fancy corp when you gonna flash your wealth asking if your cloud-patterned new mate is in need of help. However exactly 'safe-distanced' you'd be going against your pattern-based instincts. Dead. Fancy. Kind (and it is the only right standpoint inside!) Moneybag.

Blame yourself for seeking an option to kindness in making new colour-neutral, drug-neutral and sex-neutral religion when your grannies are same kind: using old dogmas.

[tomoprime]

AI 101… BUT, please view this other Mashable video (now made with real white meat) before proceeding… What the?