Connecting Dots: The Past, Present and Future of Software in an AI World | Basil Fateen | TEDxDabouq — Transcript

Full transcript

1. 0:10Software developers are weird.
2. 0:15I should know. I'm one of them. And I
3. 0:19didn't want it to be this way. I had no
4. 0:21hand in the matter. It is purely
5. 0:23genetic.
6. 0:26I'm Egyptian. And my father was one of
7. 0:29the early Egyptian computer science and
8. 0:31engineering students to work with
9. 0:32computers when they first entered the
10. 0:34country in the 60s and 70s. Now when I
11. 0:36say computers, they're not computers
12. 0:38like me and you know today. They were
13. 0:41mainframes that filled an entire room.
14. 0:44And there were two parts to this
15. 0:45mainframe. One part is where you would
16. 0:47write assembly language code, which is
17. 0:49just one level above machine code, and
18. 0:52it would put out a punch card with that
19. 0:54code. And then in the other part of the
20. 0:55main frame is where you put in the punch
21. 0:57card and it runs the code and it gives
22. 0:59you the output or tells you if there's
23. 1:00an error. Now when this entered the
24. 1:02country, the two major universities
25. 1:05fought over the mainframe. They each
26. 1:07wanted it for their engineering students
27. 1:10and then they came to a compromise.
28. 1:13We'll split it. So one university in one
29. 1:16location got the part of the mainframe
30. 1:18where you'd write the code and it would
31. 1:19give you the punch card. and the other
32. 1:21university got the part where you put in
33. 1:22the punch card and it runs the code. So
34. 1:25there was my father in his funky bell
35. 1:28bottoms
36. 1:29and afro and handlebar mustache, writing
37. 1:33code in one university and then taking
38. 1:35the punch card and getting on a bus for
39. 1:3830 minutes to get to the other
40. 1:40university and putting in the punch card
41. 1:42and praying to God that it runs
42. 1:44correctly because if not, he's going to
43. 1:46get back on the bus for 30 minutes and
44. 1:49over and over again. And apparently this
45. 1:52was such a fun and joyous experience for
46. 1:54him that he wanted the same for his only
47. 1:57son. And when I was about eight years
48. 2:00old, he pulled me away from whatever fun
49. 2:02childhood experience I was having and
50. 2:05decided to teach me how to code.
51. 2:08Now by that time it wasn't assembly
52. 2:11language. There were newer languages
53. 2:12like basic which was many levels more of
54. 2:14abstraction and simplicity higher. Of
55. 2:16course, by today's standards, extremely
56. 2:18archaic, but back then, it was at least
57. 2:21simple enough to teach a young
58. 2:23eight-year-old the basics of computer
59. 2:25science. But I rebelled instantly. I
60. 2:28told him, "I don't want to sit at a
61. 2:30computer all day like a nerd."
62. 2:33So, my dad sighed and he said, "Well,
63. 2:35what do you want to do when you grow
64. 2:37up?" And without hesitation, I told him
65. 2:39two things. One, I want to be a wrestler
66. 2:42in the World Wrestling Federation like
67. 2:44Hulk Hogan. And two, I want to write
68. 2:48stories like Rald Dah. And because the
69. 2:50second thing was slightly less insane
70. 2:52than the first, he supported me. And he
71. 2:56bought me as many books as I wanted and
72. 2:58encouraged me to write as much as I
73. 2:59could. And I totally forgot about
74. 3:01programming until I got to high school.
75. 3:04And then there was a mandatory
76. 3:05programming class. And at that time
77. 3:08there were newer programming languages
78. 3:10that were much easier than basic. But
79. 3:12despite my least efforts, somehow I
80. 3:16managed to get one of the highest grades
81. 3:17in class. And I remember as I was
82. 3:20looking at my test result, I thought,
83. 3:22"Oh, damn. Am I a nerd? I don't feel
84. 3:26like a nerd." Because at that point, I
85. 3:29was just getting into electronic music
86. 3:30and raves. And these two things did not
87. 3:32align. They did not align. So, I hid
88. 3:35that result as if I had gotten an F. I
89. 3:37didn't want anyone to know I was good at
90. 3:38this until I got to university. And then
91. 3:41my father asked me, "What major do you
92. 3:44want?" And I told him two things. I'm
93. 3:46going to major in philosophy and
94. 3:48writing. And by the way, it doesn't
95. 3:50matter much because anyway, I'm going to
96. 3:52become a famous DJ and producer like
97. 3:54Fatboy Slim.
98. 3:56And my father said, "Okay.
99. 3:59Um, well, you can always sit under a
100. 4:02tree and think about life and write. You
101. 4:04don't need a degree for that. And in
102. 4:06case the dejing doesn't work out, how
103. 4:09about you get a computer science degree
104. 4:11just as a fallback, which was a good
105. 4:13point. And then he made another good
106. 4:14point, which is that he pays for my
107. 4:16tuition. So, we both agreed that I would
108. 4:19take the computer science major
109. 4:22begrudgingly. But I knew that I was
110. 4:24going to become a famous DJ and producer
111. 4:26and be cool. Then a few years later, I
112. 4:29was up late working in my beats
113. 4:31laboratory, making sick beats and
114. 4:34cutting up samples. And then in the
115. 4:36music production software I was using,
116. 4:38it included plugins. And there was one
117. 4:40plugin I couldn't get it to sound like
118. 4:42how I wanted. And then I realized that
119. 4:44these plugins come with script files and
120. 4:46you can access them and make changes if
121. 4:48you could. And I opened up the script
122. 4:49file and I realized I know this
123. 4:51programming language. So I was able to
124. 4:54make the changes. And then after I made
125. 4:57the changes, I realized, well, wouldn't
126. 4:58it be cool if it does this? And I
127. 5:00started to add more and more to the
128. 5:01plug-in file until it became three times
129. 5:04the size. and had 10 times more
130. 5:06features. And about 10 hours later, I
131. 5:08stopped horrified at what's happening.
132. 5:10And I said, "Oh, damn."
133. 5:12And then I begrudgingly went a bit
134. 5:15deeper into the software rabbit hole,
135. 5:17but I knew this is still temporary. One
136. 5:19day I'm going to be cool.
137. 5:21And then I got to become a senior
138. 5:24software developer in a major software
139. 5:26house handling a very complex and big
140. 5:29project. And I was there late one night.
141. 5:32Everyone had gone. And I had just
142. 5:35started checking the log files before I
143. 5:36go. I put on my jacket and my wife calls
144. 5:38me saying, "Hey, just checking when are
145. 5:40you coming home so that we can plan for
146. 5:42dinner." And I told her and just as I
147. 5:45was talking to her, I realized there's a
148. 5:47bug in the logs. And a bug is something
149. 5:51that goes wrong in code. Now, there's
150. 5:53two kinds of bugs. There's the kind of
151. 5:55bug that's pretty simple and you could
152. 5:57fix it quite quickly. And then there's
153. 5:59the second kind of bug which takes you
154. 6:01into the depths of hell and to the
155. 6:03brinks of your own sanity. This was the
156. 6:06second kind, but I didn't know that at
157. 6:08that point. And I told her, "Yeah, I'm
158. 6:10just going to be probably home in about
159. 6:1215 minutes. I just need to check this."
160. 6:13And she said, "Great, because I'm making
161. 6:15your favorite sweet and sour chicken."
162. 6:17And I told her, "That sounds amazing,
163. 6:18baby. I love you. I'll see you in 15
164. 6:20minutes." And I started checking the
165. 6:22logs and doing a bit of traces, and
166. 6:25everything seemed correct. I don't
167. 6:27understand what's going on. Everything
168. 6:28is running correctly. And then I waited
169. 6:31for the error to happen and it didn't.
170. 6:32So I put my jacket back on about to
171. 6:34leave. The error happens again. How is
172. 6:37this possible? Why is it random? Why is
173. 6:40it not happening every time at a certain
174. 6:42situation? And then every 15 minutes, my
175. 6:45wife would send me a message. Where are
176. 6:47you? When you coming home? And I tell
177. 6:48her, just 15 more minutes. I I think I
178. 6:50know what it is. Just 15 more. I'll be
179. 6:52there in 15 minutes. An hour passes. 3
180. 6:54hours. 5 hours. My wife sends me one
181. 6:57final SMS containing one single emoji,
182. 7:02and I can't tell you what that emoji is,
183. 7:05but it's not one of the good ones. And I
184. 7:08did not have sweet and sour chicken that
185. 7:10evening.
186. 7:12And I tried to figure out what the bug
187. 7:14was, and I couldn't for three days.
188. 7:19By the day three, I was a shell of the
189. 7:22man I was on day one because I would
190. 7:24stay up so late, not figure it out, go
191. 7:27home, sleep, dream of the code, wake up
192. 7:29thinking I had figured it out, go back
193. 7:31to work early, still hunt for the bug.
194. 7:34My wife stopped sending me messages and
195. 7:36I think she started to communicate with
196. 7:38a divorce lawyer at the time.
197. 7:40But on the evening of the third night, I
198. 7:43thought of something and it was
199. 7:45something I learned in university, but I
200. 7:47had never really seen it in the field.
201. 7:49There's this thing called a race
202. 7:50condition which happens randomly when
203. 7:53different parts of the system are
204. 7:54inserting and reading from a table and
205. 7:56you're expecting them to have a certain
206. 7:58sequence but for whatever reason one of
207. 8:00them becomes a bit slow and the sequence
208. 8:02changes and this is a very rare thing
209. 8:04but it is known. So I made the necessary
210. 8:07adjustments and I put the logs and I
211. 8:10waited
212. 8:121 hour, two hours, three hours the bug
213. 8:15didn't happen. I finally figured out the
214. 8:18bug and I let out such a massive scream
215. 8:21of joy that the security in the ground
216. 8:24floor came up to my floor with guns
217. 8:26drawn thinking I was being brutally
218. 8:28murdered. But I didn't care.
219. 8:32I can't make the sound I made but it
220. 8:35looked a little bit like this.
221. 8:38Yeah.
222. 8:44Those three days culminating in that
223. 8:46moment was maybe some of the most joy
224. 8:50I've ever felt in my life. Possibly tied
225. 8:53with the birth of my first child and far
226. 8:56ahead of the birth of my second.
227. 8:59And it was at that moment that I
228. 9:01realized two things. First, I am a
229. 9:05software developer and I love it and
230. 9:07there's nothing cooler. And the second
231. 9:10thing is I was looking for role models
232. 9:12all around without realizing I had a
233. 9:14great professional role model at home.
234. 9:16And I called my dad right then and I
235. 9:18told him this. And there was a few
236. 9:21seconds of silence. And then he said,
237. 9:32>> [applause]
238. 9:37[applause]
239. 9:38>> Sorry.
240. 9:42So, like I said, software developers are
241. 9:45weird.
242. 9:47We love to solve complex problems and
243. 9:50figure out ways to create better
244. 9:52solutions. It's for most of us be beyond
245. 9:55just a job. It's something we find true
246. 9:58meaning in. But things are changing and
247. 10:03it's due to generative AI and we're all
248. 10:06in this transformational wave right now
249. 10:08and it's growing exponentially
250. 10:11and right now I'm the principal tech
251. 10:13evangelist at Amazon Web Services AWS.
252. 10:16So I have a very specific view on the
253. 10:19paradox that's unfolding because on one
254. 10:22hand
255. 10:23there's never been a better time to be
256. 10:25an innovator because it's never been
257. 10:27easier to bring your idea to life. On
258. 10:30the other hand,
259. 10:32things are happening so quickly that
260. 10:34it's very disorienting and people are
261. 10:37unable to quite figure out with this
262. 10:39rate of change where do they fit in and
263. 10:41where does AI fit in especially with
264. 10:42software and AI. But to truly understand
265. 10:46how these two worlds are merging, we
266. 10:49need to go back to the 9th century but
267. 10:51stay in the Middle East.
268. 10:53And our ancient scholars weren't just
269. 10:55extremely smart. Many of them were
270. 10:58polymaths. And what a polymath is is
271. 11:00someone who has expertise in a variety
272. 11:02of fields. And this allowed them to
273. 11:04connect the dots between astronomy,
274. 11:07philosophy, art, music, medicine, and
275. 11:10create truly groundbreaking
276. 11:12breakthroughs. And maybe one of the most
277. 11:14notable is Alarismi
278. 11:16who wrote a text in the 9th century that
279. 11:19first introduced the concept of the
280. 11:21algorithm. And an algorithm is the set
281. 11:24of steps needed to solve a problem. And
282. 11:26it's still one of the fundamental
283. 11:27concepts of computer science today. And
284. 11:30now if we go to the 20th century, from
285. 11:33east to west, from a big beard to mutton
286. 11:35chops, we go to Isaac Azimoff. He was
287. 11:38one of the most prolific science fiction
288. 11:40authors to ever exist. And in the 1940s,
289. 11:43his stories about worlds where robots
290. 11:46and AI coexist with humans and the
291. 11:48implications of this inspired an entire
292. 11:51generation of thinkers, artists, and
293. 11:53scientists, including it is said the
294. 11:56people who attended the Dartmouth
295. 11:59conference. Now, these brainiacs got
296. 12:01together to talk about how can we make
297. 12:05machines think, but then they realize
298. 12:07they first have to ask another question.
299. 12:09How do we think? And out of those
300. 12:12conversations came groundbreaking AI
301. 12:14research into natural language
302. 12:15processing and neural networks.
303. 12:18Then in 1997,
304. 12:20we watched our chess grandmaster
305. 12:22champion Gary Kasparov go up against
306. 12:25IBM's Deep Blue, a very rudimentary form
307. 12:28of AI that's only trained to play chess.
308. 12:31And the entire world was shocked to
309. 12:32watch our champion lose to the machine
310. 12:36at chess, the thinking man's game.
311. 12:39In the 2010s, we started to see machine
312. 12:41learning improve the personalized
313. 12:44recommendations on sites like Amazon and
314. 12:46different streaming sites and we stopped
315. 12:47getting spam in our email inbox because
316. 12:49it was able to filter it. But only when
317. 12:51four elements converged did we get
318. 12:53generative AI as we understand it and it
319. 12:56started with a new algorithm over a
320. 12:58century after Alawarismi introduced the
321. 13:01concept of the algorithm and this new
322. 13:03algorithm was the transformer model. the
323. 13:06transformers, the T in GPT. And this
324. 13:09transformer model was a new algorithm in
325. 13:10the world of AI that allowed the AI to
326. 13:13not just predict and classify based on
327. 13:15data, but generate entirely new data.
328. 13:18But this algorithm needed an enormous
329. 13:20amount of public data. But to be able to
330. 13:22train that public data through this
331. 13:24transformer model, you needed dedicated
332. 13:26processors and very powerful GPUs. And
333. 13:29to do that in a scalable method, you
334. 13:31needed cloud computing. And this is how
335. 13:33these four elements came together. And
336. 13:35this is how generative AI came to be and
337. 13:38was born. And now it's fundamentally
338. 13:42changing software development and the
339. 13:44nature of software developers because
340. 13:47now you can use natural language,
341. 13:50English sentences to generate code. Once
342. 13:54again going many more levels of
343. 13:56abstraction higher than the days when my
344. 13:58dad was writing assembly code on those
345. 14:00punch cards. Generative AI can write
346. 14:03code, explain code, debug code, document
347. 14:06code, port code from one language to
348. 14:08another. It can build a front end. It
349. 14:10could build the back end. It could
350. 14:11connect the two. So then it begs the
351. 14:14question, if generative AI can do so
352. 14:16many of those tasks that traditionally
353. 14:19software developers do, then what's left
354. 14:21for software developers to do? In other
355. 14:24words, what do devs do when they don't
356. 14:26dev? And is it still worth learning how
357. 14:31to code?
358. 14:34I believe yes, 100%.
359. 14:38And here's why. It's because
360. 14:41developers are weird. And it's precisely
361. 14:45because of our strange, irrational,
362. 14:49unpredictable paths. going after things
363. 14:51that are exciting to us, chasing these
364. 14:53curiosities of developing solutions,
365. 14:56struggling for days over complex
366. 14:58problems. It's that lived experience
367. 15:01that teaches us how to connect the right
368. 15:03dots for future software projects. And
369. 15:07this lived experience is where we build
370. 15:09our unique expertise. And this lived
371. 15:12experience is not public data and it's
372. 15:15not accessible by generative AI.
373. 15:18Therefore,
374. 15:20a software developer who knows what
375. 15:22they're doing plus generative AI has
376. 15:24such a massive edge over just generative
377. 15:28AI.
378. 15:30And it's due to two things, context and
379. 15:33complexity. Because software development
380. 15:35is not just coding. Generative AI can
381. 15:38produce code fast, sure, but it lacks
382. 15:41the true contextual understanding
383. 15:43between mediocre code and excellent
384. 15:46software that scales. AI can create a
385. 15:50simple prototype, sure, but the more
386. 15:52complex the project gets, the more it
387. 15:53will add technical debt and likely cause
388. 15:56a lot of future heart attacks when
389. 15:58things go wrong and we don't understand
390. 15:59what happened. AI can generate a
391. 16:02solution, but it doesn't understand if
392. 16:04it's the best solution for this
393. 16:05particular team in this new business use
394. 16:09case. Context helps us connect the right
395. 16:12dots. And it's the difference between
396. 16:14intelligence and wisdom.
397. 16:16And I've been thinking about this not
398. 16:19just for the work I do with current
399. 16:21developers, but for future developers.
400. 16:24Because right now, Luly, my daughter, is
401. 16:2810 years old and I believe that 100%
402. 16:32this is the time to start coding. And I
403. 16:34decided I'm going to teach her to start
404. 16:37coding. And the day I decided, I had
405. 16:41butterflies in my stomach. And as I
406. 16:44walked up the steps to teach her, I was
407. 16:46filled with emotion. I was hearing
408. 16:48sirens. I felt like I'm part of this
409. 16:51cosmic grand story right now. And I
410. 16:54reached her room and I knocked on the
411. 16:56door and I went in and she was on her
412. 16:58iPad and I said, "Hey, Luly." And she
413. 17:00said, "Hi Bubba." And I said, "How about
414. 17:04I teach you how to code?" And she said,
415. 17:07"I know how to code."
416. 17:09And I said, "Huh?"
417. 17:12and she came over and showed me on her
418. 17:14iPad all of the gamified coding apps
419. 17:17that she's been learning, all of the
420. 17:18fundamentals of programming, computer
421. 17:20science, and the generative AI threads,
422. 17:23creating a website and a business plan
423. 17:26for that website. And I just stood there
424. 17:29staring at her. I said, "Oh, damn.
425. 17:34Thank you.
426. 17:36[applause]