Artificial intelligence (AI) uses computers to solve problems or make automated decisions typically requiring human intelligence. The two major AI techniques are rules-based and machine learning approaches. Rules-based AI uses logical rules to automate processes, while machine learning algorithms find patterns in data to improve performance over time without being explicitly programmed. Today, AI is mostly "weak" and pattern-based, not capable of human-level reasoning, but it has automated many tasks through proxies like statistical patterns. Hybrid systems combining approaches work best.

1. ArtificialIntelligence
Overview
Mithileysh Sathiyanarayanan

2. Artificial IntelligenceOverview
1.What is Artificial Intelligence?
2.Major Artificial IntelligenceTechniques
• Rulesand LogicBasedApproach
• Machine Learning BasedApproach
• Hybrid System
3.Limits of Artificial IntelligenceToday

3. What is
Artificial
Intelligence?

4. Artificial Intelligence (AI)
• What is Artificial Intelligence(AI)?
• Using computers to solveproblems
• Or make automated decisions
• For tasks that, when done byhumans,
• Typically require intelligence

5. Limits of ArtificialIntelligence
• “Strong”ArtificialIntelligence
• Computers thinking at alevel that meets or surpassespeople
• Computers engaging in abstract reasoning & thinking
• Thisisnot what we have today
• Thereisnoevidencethat we are closeto Strong AI
• “Weak”Pattern-Based Artificial Intelligence
• Computers solve problems by detecting useful patterns
• Pattern-based AI is an Extremely powerful tool
• Hasbeen used to automate many processestoday
• Driving, language translation
• Thisis the dominant mode of AItoday
✔
✘

6. Major AIApproaches
TwoMajor AITechniques
• Logic and Rules-BasedApproach
• Machine Learning (Pattern-BasedApproach)

7. Logic and Rules-BasedAI

8. Logic and Rules-BasedApproach
• Logic and Rules-BasedApproach
• Representing processesor systemsusing logical rules
• Top-down rules are created for computer
• Computers reason about thoserules
• Canbe usedto automateprocesses

9. Machine Learning

10. Machine Learning (Patternbased)
• Machine Learning(ML)
• Algorithms find patterns in data and infer rules on their own
• ”Learn” from data and improve overtime
• Thesepatterns canbe usedfor automation orprediction
• MLis the dominant mode ofAI today

11. Machine LearningUses
Self-Driving Vehicles Automated
recommendations
Computer
Translation

12. Machine Learning Main Points
Learning
Pattern Detection
Data
Self-Programming

13. Spam or Wanted Email?
System detects patterns in Email
About likely markers of spam
Detected Pattern
Emails with “Earn Cash”
More likely to be spam email
Can use such detected patterns to
make automated decisions about
future emails
Example: Email Spam Filter
“EarnCash”
“EarnCash”
detected
in 10% of Spamemails
0%of wantedemails

14. Identification Improves
Algorithm improves in performance
In auto-identifying spam
As it is able to examine more data
And find additional indicia of spam
Algorithm is “learning” over time
from additional examples
Example: Email Spam Filter
“Free”
Probability of Spam
Contains
“Free”
70%Spam
Contains
“Earn Cash”
90%Spam
From
Belarus
85%Spam

15. Intelligent Results Without
Intelligence
For some (not all) complex tasks
Requiring intelligence
Can get “intelligent” automated
results without intelligence
By finding suitable
Proxies or Patterns

16. Proxies for Intelligent Results
Without Intelligence
Statistical Machine Translation
People use advanced cognitive skills to
translate
Google finds statistical correlations by
analyzing previously translated
documents
Produces automated translations using
statistical likelihood as
a “proxy” for underlying meaning

17. Proxy Principle for Automation
Detecting
Patterns
That can serve as
Proxies
For some underlying
Cognitive Task

18. Machine Learning Main Points
Learning
Pattern Detection
Data
Self-Programming

19. Summary Major AIApproaches
TwoMajor AITechniques
• Logic and Rules-BasedApproach
• Machine Learning (Pattern-BasedApproach)

20. Hybrid Systems
• Many successfulAI systemsare hybrids of
• Machine learning & Rules-BasedHybrids
• e.g. Self-driving carsemploy both approaches
• Human intelligence +AI Hybrids
• Also, many successfulAI systemswork best when
• Theywork with humanintelligence
• AI systemssupply information for humans

21. Humans
+
Computers
Technology Enhancing
(Not Replacing) Humans
>
Humans Alone
ComputersAlone

22. Examples of AI in LawToday
• Machine Learning
• AI in Litigation - E-Discovery and ”PredictiveCoding”
• Natural LanguageProcessing(NLP)of Legal Documents
• Automated contractanalysis
• Predictive Analytics for Litigation
• Machine LearningAssisted LegalResearch
• Logic and Rules-BasedApproaches
• Compliance Engines
• Expert Systems
• Attorney Workflow RuleSystems
• Automated DocumentAssembly

23. Limits on ArtificialIntelligence
• Artificial IntelligenceAccomplishments
• Automate many things that couldn’tdo before
• Limits
• Many things still beyond the realm of AI
• No thinking computers
• NoAbstract Reasoning
• Often AI systemsHaveAccuracy Limits
• Many things difficult tocapture in data
• Sometimes Hard to interpretSystems

24. Questions?