SiriusLabs A.I was developed in the 1950s by 5 scientists with the following objectives a. Reasoning b. Knowledge Representation c. Navigation d. Natural Language Processing e. Perception & Depth f. General Intelligence
to get Computers to reason like us and do the complex things we did. Like solve algebra problems and diagnose disease. The overall goal was to be presented with a problem and then the computer reasoned it’s way to the answer. Deep Blue’s first reasoning prototype game.
machines context for them to be able to fully understand things. Teaching them about the world and objects around. Funny enough, just the same way this is the way babies learn, which is why I think they stare a lot. A baby is wondering when he can make out what a TV is, “What the heck is this shit? What the heck is that?” Creator of Lisp, John McCarthy, which was a programming language invented just to teach computers about objects, languages and the world.
computers how to move and plan navigation around our world. Telling them which paths are safe, which paths are optimal and which paths lead to which goals. The first self-moving robot/computer developed at SRI in Menlo Park, California in the mid 1960s. If you look closely, you see camera’s and sensors to help guide it’s movements.
This ties into point 2 and describes we teaching computers and machines how to recognise our communication to them, through voice, text or even sign language like a normal human being would. And also how to communicate back to us in the same medium. Understanding also the semantics. *AI->WInter Use the “the spirit is willing but the flesh is weak“
We tried to train computers through a set of eyes(video feeds) a set of data and how to make sense of that data. To understand why is a ball on top of a plate etc. To categorise and identity what is what. Same thing done though a set of touch, smell and hearing.
last of the plan of the founding fathers was the hope that teaching computers all these above would lead to generalized intelligence that was hard to program in or originally not programmed in, things like intuition, creativity, humour, emotional intelligence etc
as far back as 1950s, you are tempted to ask, how come we don’t have significant advancements now? Like Jarvis and Friday from Iron man? The answer to that would be the period of the AI Winter. Where the technology was too expensive to warrant further investments based on the previous results it had already produced. *Life Coach Example. Using ELIZA *NLP Example. Converting Russian to English and back. 2. A.I Winter
made recently that gave rise to the current wave of AI and Machine Learning. It was Deep Learning. While the other techniques listed above tried to treat human expert knowledge as set of rules to feed a computer. Deep learning instead created a set of data structures based on a given set of data, feed that data structure to the computer and let the computer by itself figure things out based on these models or data structures. And it worked better and inexpensive this time due to scale, better investment, better algorithms. Especially when Google stepped in and gave Andrew Ng the resources to scan 10m youtube videos. 2. A.I Winter
a. Reasoning – AlphaGo. There are more combinations in AlphaGo than there are of atoms in the Universe. AlphaGo helped us understand that. b. Knowledge Representation – It’s easier for computer to easily diagnose diseases based on samples collected. Things like which red blood cells have cancer, if an irregular heartbeat means stroke, like in Apple’s Watch. Mention Gyroscope App + Fitbit integration. c. Navigation – Fully autonomous cars are now making their ways into the market. Mention Voyage and Udacity’s self-driving program. George Holtz and Comma.ai d. Natural Language Processing – Alexa. Google’s speech to text library. Anger, Emotions. Siri
e. Perception & Depth – Autonomous cars now can identity objects, moving cars or human beings. Autonomous camera drones can also identify target objects and focus. I feel it won’t be long before we start to see autonomous kill drones f. General Intelligence – DeepJazz, a library that makes jazz on the fly based on analysed samples. Available on GitHub. Google’s Neural Network paintings
Narrow A.I (A.I doing a specific task) -> Where we are now 2. General A.I (A.I has smart as humans) 3. Super A.I (A.I with intellect higher than humans). For the paranoid, A.I still has a long way to go to successfully achieve general A.I. Because Narrow A.I where we still are, is still very limited. Mention an umbrella up or down.
all this? AI is going to get into all software, the same way database got into software. So it means that in a couple of years, saying your software is powered by A.I would be the same thing as saying we have a database. Airways: Uses drone footage to analyse if mining rock road barriers are tall enough. Everlaw: They help lawyer prepare adequately for trials. By sampling a lot of other similar trails and results. Know which documents to focus on and which to hand over to the counsel. Airbnb: To tell you the best recommended prices at the right times.
all this? Apple Watch: Uses heart beat to predict diseases, cancer, sleep problems or high blood pressure Zipline: In rwanda uses machine learning to determine optimal routes to pick and deliever blood to receipents Pinterest: Identifing bags and clothes in a picture. Instacart: Telling you the right aisle and section to go to get your groceries.
all this in fintech? 1. Process Automation. Take all the manual work and automate it. * Chatbots * Call-center automation. * Paperwork automation. – Reviewing documents. Scanning through. Approval requests. * Gamification of employee training, and more. 2. Security. Detecting and reducing fraud. A lot of ATM theft that happen in this country, I believe can benefit from this. A quick scan of the user at the machine, then send an approval request to the owner of the ATM. 3. Underwriting and creditscoring: See how credible a customer is against past loans, spendings, money management
all this in fintech? 4. Algorithm trading. Help predict and make better trading decisions. Such as the best time to sell or buy certain stocks 5. Robo-advisory. You set a financial goal and it helps you optimise your assets to hit that goal. Making wise investments. Also in terms of personalized insurance. 6. Personalized savings offerings: say a savings fintech startup has a user who puts in money regularly at the 25th of each month and one month they don’t (maybe due to job loss), you can use that to flag off and send a personalized offering for example, for the next two months, we are giving you a 12% savings interest on your already saved money as against the orignal 7% because we understand.
identify how well Machine Learning can work across your company or organisation. And also in your products. It might just be helping you automate a usually tedious repetitive process which then on can free your team to do more creative process. Mention Artemis’ data visualisation and machine learning. After identifying, you can buy already made machine learning products from vendors that suite your requests, you can have a custom machine learning model built for you and you can build your own of the back of existing APIs and libraries.
your team, to understand the offerings of A.I and machine learning and learn to embrace. There are a lot of cheap Udacity and coursera courses that can get you started.
genda
usanchuu
ezaki
makocchi
kzkhykw1991
kenichiodo
terapyon
endonanana
nttcom
ishikawa_satoru
akhino
odashi
aleyda
michielstock
minecr
chriscoyier
malarkey
codingconduct
notwaldorf
ipullrank
coloredviolet
letsgokoyo
aki_iinuma
nonsquared
