AIFactMyth2.pdf

Transcript

2. DJ Daugherty dj@augustwenty.com
3. not an expert! NLPs, LLMs or GenAI
4. not against AI! just a different perspective
7. think.
10. How should we be thinking about AI?
11. How should we be thinking about AI? As a business that solves technology problems
12. myths of artificial intelligence
13. Myth AI is synonymous with human-like intelligence. Myth #1
14. Myth AI is unbiased and objective. Myth #2
15. Myth AI is only accessible to large corporations and tech giants. Myth #3
16. realities of artificial intelligence
17. Power to access volumes of knowledge instantaneously.
18. To create fully realized illusory other worlds
19. To deceive, to conjure, to transport
20. To materialize on a massive scale
21. These powers are magical!
22. my concern!
23. not clear its that humanity has the
24. Tools or framework… not clear its that humanity has the
25. Tools or framework… use them responsibly! not clear its that humanity has the
26. What’s at stake?
27. What’s at stake?
28. 1945 clear to the creators of the atom bomb!
29. “Now, I am become Death, the destroyer of worlds.” — J. Robert Oppenheimer
30. “It is a profound and necessary truth that the deep things in science are not found because they are useful; they are found because it was possible to find them.” — J. Robert Oppenheimer
31. “It is a profound and necessary truth that the deep things in science are not found because they are useful; they are found because it was possible to find them.” — J. Robert Oppenheimer
32. “It is a profound and necessary truth that the deep things in science are not found because they are useful; they are found because it was possible to find them.” — J. Robert Oppenheimer
33. Perverse instantiation
35. this is not just an AI problem.
36. we have to be careful.
37. We are messing with potentially world altering powers
38. Technology enables us to deceive on a massive scale
39. It’s magic
40. It’s magic because it involves the manipulation of reality
41. elections
42. TSA
43. phone call in your voice
44. we should look at it differently now that we understand that there is world altering magic at play…
45. Seed example
46. why so serious?
47. addicted… relentless forward progress
48. we celebrate it! venture capital pours billions into it
49. its not all bad! as we become more technologically sophisticated… we will be able to capture, harness and apply power to alleviate suffering
50. better shelter
51. better shelter better food
52. better shelter better food better medicine
53. but…
54. but… let’s revisit the concern.
55. home
56. home nuclear
57. home nuclear weapons
58. home nuclear weapons kit
59. home nuclear weapons kit good coder… working on something nobody knows about
60. home nuclear weapons kit good coder… working on something nobody knows about potentially unleashing something that is world altering
61. we don’t hand people nuclear weapons why are we handing AI to everyone?
62. the destination is not the addictive part… its the relentless forward drive
63. the destination is not the addictive part… its the relentless forward drive AI powered RPi project
64. the destination is not the addictive part… its the relentless forward drive AI powered RPi project home nuclear weapons kit
65. numerous papers that claim that AI will solve all of the worlds big problems
66. Climate crisis / world hunger We already know how to solve these issues… We have to change
67. Biggest fact! Everything has AI now!
68. AI is a real field of computer science studied for decades… and recently has made major strides
69. That’s real… let’s talk about the marketing term… AI
70. Metaverse
71. Metaverse augmented reality
72. Metaverse augmented reality web3
73. Metaverse augmented reality web3 cryto
74. Metaverse augmented reality web3 cryto blockchain
75. Metaverse augmented reality web3 cryto blockchain and now! AI
76. Spotify released a new ‘AI DJ’ to generate custom playlists Funny! It already did that
77. how many apps do you have that are now AI powered?
78. Remember self driving cars $100 billion - self driving cars are going nowhere
79. generative AI large amounts of data… text… images… videos. it can output a similar result using probability… e.g. which word… pixel… video frame is next
80. Incredibly sophisticated imitation systems e.g. if we give it every Dr. Seuss story… it can output a new Dr. Seuss story very similar to every other Dr. Seuss story
81. In the town of Columbus, so lively and bright, There's a tale to be told, under stars in the night. It's a story of DJ Doherty, you see, And an AI conference, as grand as can be. It was set for the fourteenth, in the month of May, A gathering where tech minds would dance and sway. The topic was "AI, a Rising Sun," And the theme was a miss and mystery, all spun. Now, DJ Doherty, a tech wiz so keen, Had gadgets and gizmos that few had seen. With a twirl of his knob and a press of a key, He'd make robots dance and screens light up with glee. The conference was held at the big town hall, With posters of circuits that covered the wall. People from near, people from far, Came to see AI, the rising star. The mystery part, oh, it was grand, With puzzles and riddles right at hand. "Which bot can think, which one can feel?" But alas, poor DJ, in his rush and his run, Forgot one little thing that weighed a ton. His magical, mystical, mystery box, Stuffed with an AI that outsmarts a fox. Without his box, at half past two, DJ felt lost, not knowing what to do. But the crowd they cheered, "Come on, don't frown!" "For in this miss, no need to drown." So DJ thought, and DJ spun, A tale of AI that had just begun. With words so bold, and ideas so bright, He turned his miss into the highlight of the night. The crowd was amazed, intrigued, and more, With stories of AI lore they adore. The conference a hit, a mystery no more, All thanks to DJ, who the crowd did adore. And so it goes in the land of code and wire, A story of a miss that turned into fi re. For in Columbus, on that magical day, The mystery was solved in an unforgettable way.
82. Expand to the internet!
83. Language models don’t work this way… they only predict ‘what’s next’ based on existing data Because the training data is the entire internet… that means lots of ‘probably not good’ data
84. When a regular search engine gives you crap… you use your brain humans are good at differentiating between ‘Joe Joe the Circus Clown’ and the Mayo clinic
85. Generative AI is a parrot The problem is… people believe it!
86. Predicted this phenononom
87. Humans are hard-wired for language If we see a string of words that make grammatical sense we naturally search for and find meaning in it and we naturally tend to assume there must be a mind like ours behind it
88. Even when… what we are looking at is word salad barfed out by a problematic parrot. The researchers wrote, “the tendency of human interlocutors to impute meaning where there is none can mislead both NLP researchers and the general public into taking synthetic text as meaningful”
89. The researchers also wrote, “Combined with the ability of LLMs to pick up on both subtle biases and overtly abusive language patterns in training data, this leads to risks of harms, including encountering derogatory language and experiencing discrimination at the hands of others who reproduce racist, sexist, extremist or other harmful ideologies reinforced through interactions with synthetic language.”
90. The authors stated, “We call on the field to recognize that applications that aim to believably mimic humans bring risk of extreme harm” Published paper deemed biased. Google fired her!
91. Like firing the weather person because you do not like the weather
92. Even the name… AI, is misleading AI wouldn’t function without humans.
93. Manually removing content Sludge/toxic content Training data is human generated
94. a couple funny stories…
97. Create new things, think new thoughts, discover new facts
98. AI is not a human mind… not anything like a human mind it just imitates the past… based on probability
99. The human brain has evolved over many generations… it is so complex we don’t even understand how it works, but we use this marvelous meat machine in our skulls to observe the world around us
100. We reason why it is the way it is… we communicate w/ other minds what we have learned and we create new things… and we used it to create chatgpt… an algorithm to create Dr. Seuss stories.
101. AI = Artificial Imitation
102. Myth AI is synonymous with human-like intelligence. Reality While AI has made remarkable advancements, it still lacks the depth and breadth of human intelligence. Current AI systems excel in narrow tasks and specific domains but struggle with tasks that require common sense reasoning, creativity, and emotional understanding. AI operates based on algorithms and data, devoid of consciousness and true understanding. Myth #1
103. Myth AI is synonymous with human-like intelligence. Reality While AI has made remarkable advancements, it still lacks the depth and breadth of human intelligence. Current AI systems excel in narrow tasks and specific domains but struggle with tasks that require common sense reasoning, creativity, and emotional understanding. AI operates based on algorithms and data, devoid of consciousness and true understanding. Myth #1
104. Myth AI is unbiased and objective. Reality AI systems are prone to inheriting biases present in the data they are trained on, reflecting societal prejudices and inequalities. Without careful oversight and mitigation strategies, AI can perpetuate discrimination and reinforce existing social biases. Addressing bias in AI requires a concerted effort to ensure fairness, transparency, and accountability throughout the development lifecycle. Myth #2
105. Myth AI is unbiased and objective. Reality AI systems are prone to inheriting biases present in the data they are trained on, reflecting societal prejudices and inequalities. Without careful oversight and mitigation strategies, AI can perpetuate discrimination and reinforce existing social biases. Addressing bias in AI requires a concerted effort to ensure fairness, transparency, and accountability throughout the development lifecycle. Myth #2
106. Myth AI is only accessible to large corporations and tech giants. Reality The democratization of AI is underway, with a proliferation of open-source tools, platforms, and educational resources. Startups, small businesses, researchers, and hobbyists have access to AI technologies, enabling innovation across diverse sectors. Cloud-based AI services further lower barriers to entry, allowing organizations of all sizes to leverage AI capabilities without significant upfront investment. Myth #3
107. Myth AI is only accessible to large corporations and tech giants. Reality The democratization of AI is underway, with a proliferation of open-source tools, platforms, and educational resources. Startups, small businesses, researchers, and hobbyists have access to AI technologies, enabling innovation across diverse sectors. Cloud-based AI services further lower barriers to entry, allowing organizations of all sizes to leverage AI capabilities without significant upfront investment. Myth #3
108. Myth AI should be openly used by developers on all projects. Myth #4
109. Myth AI should be openly used by developers on all projects. Myth #4
110. what do I want you to take away?
111. recognize that AI (LLMs) are an imitation
112. AI is largely a marketing term
113. think about how you are leveraging AI to create an advantage for your business and/ or partners
114. DJ Daugherty, co-founder/CTO dj@augustwenty.com
115. the end