1
00:00:01,320 --> 00:00:04,760
Speaker 1: Now, artificial intelligence is rapidly reshaping the world of research,

2
00:00:04,960 --> 00:00:08,319
Speaker 1: changing how ideas are developed, tested and shed. What was

3
00:00:08,360 --> 00:00:11,720
Speaker 1: once a slow, manual process is now being accelerated by

4
00:00:11,760 --> 00:00:16,200
Speaker 1: tools that can analyze vast amounts of data, summarize complex information,

5
00:00:16,360 --> 00:00:19,759
Speaker 1: and even suggest new directions for study. The shift is

6
00:00:19,800 --> 00:00:23,160
Speaker 1: not just about speed, but about unlocking entirely new ways

7
00:00:23,160 --> 00:00:27,320
Speaker 1: of thinking and discovering, from healthcare breakthroughs to academic innovation.

8
00:00:27,640 --> 00:00:30,639
Speaker 1: AI is already playing a critical role behind the scenes.

9
00:00:30,720 --> 00:00:34,880
Speaker 1: But with this progress comes the important question around accuracy, ethics,

10
00:00:35,040 --> 00:00:37,920
Speaker 1: and the role of human judgment. As these tools become

11
00:00:38,000 --> 00:00:41,080
Speaker 1: more widely available, researchers are being challenged to adapt and

12
00:00:41,159 --> 00:00:45,199
Speaker 1: develop new skills to stay relevant. One such researcher is

13
00:00:45,240 --> 00:00:48,519
Speaker 1: doctor Samira Goliza There. She's a materials and mechanical engineer

14
00:00:48,560 --> 00:00:52,080
Speaker 1: and researcher and an AI specialist working at the intersection

15
00:00:52,159 --> 00:00:55,200
Speaker 1: of machine learning and advanced engineering systems. She's with the

16
00:00:55,320 --> 00:00:58,480
Speaker 1: University of Cape Town. Good morning, doctor Goliza.

17
00:00:58,040 --> 00:00:59,960
Speaker 2: There, Good morning Ray. How are you today.

18
00:01:00,200 --> 00:01:02,279
Speaker 1: I'm doing so good and it's a public holiday. Did

19
00:01:02,320 --> 00:01:03,840
Speaker 1: you have a good long weekend?

20
00:01:04,200 --> 00:01:06,640
Speaker 2: Yes, it was perfect. I came to the port of

21
00:01:06,880 --> 00:01:08,240
Speaker 2: such a beautiful place here.

22
00:01:08,360 --> 00:01:11,120
Speaker 1: Oh, that's a stunning place. A stunning place, Samita. You

23
00:01:11,200 --> 00:01:13,960
Speaker 1: are hosting not one, but two workshops coming up in

24
00:01:14,000 --> 00:01:17,280
Speaker 1: the next couple of weeks. Both workshops focus on the

25
00:01:17,319 --> 00:01:21,080
Speaker 1: topic of AI. One of your workshops focuses on AI

26
00:01:21,200 --> 00:01:24,120
Speaker 1: in the medical field, and then a different focuses on

27
00:01:24,240 --> 00:01:26,840
Speaker 1: AI in research, which is kind of what we're talking

28
00:01:26,880 --> 00:01:30,840
Speaker 1: about today. So to start us off, could you give

29
00:01:30,880 --> 00:01:34,200
Speaker 1: us this broad definition within your research? What do you

30
00:01:34,200 --> 00:01:36,440
Speaker 1: mean by this AI revolution in research?

31
00:01:36,920 --> 00:01:37,080
Speaker 3: Oh?

32
00:01:37,240 --> 00:01:41,440
Speaker 2: Actually, Buried, it refers to how AI is transforming every

33
00:01:41,600 --> 00:01:45,000
Speaker 2: stage of research. So we have different different stages in

34
00:01:45,000 --> 00:01:48,920
Speaker 2: our research, from generating the ideas and brainstorming to the

35
00:01:49,080 --> 00:01:51,760
Speaker 2: data analyzers, writing a paper and at the end it's

36
00:01:51,800 --> 00:01:56,360
Speaker 2: going to be a publication. It is ultimately fundamentally changing

37
00:01:56,440 --> 00:02:01,040
Speaker 2: not only how quickly but also how effectively and intelligently

38
00:02:01,160 --> 00:02:03,080
Speaker 2: knowledge is produced through the AI.

39
00:02:03,600 --> 00:02:06,440
Speaker 1: So in terms of now versus five years ago, not

40
00:02:06,520 --> 00:02:09,919
Speaker 1: even five years ago, I would say, as recent as

41
00:02:09,960 --> 00:02:13,440
Speaker 1: three years ago, could you compare just how different research

42
00:02:13,600 --> 00:02:14,840
Speaker 1: looks then to now?

43
00:02:16,040 --> 00:02:19,840
Speaker 2: Yes, If we compared just a few years ago, research

44
00:02:20,080 --> 00:02:23,800
Speaker 2: was a large largely manual and it was all depend

45
00:02:23,880 --> 00:02:28,519
Speaker 2: on the researchers, and it was absolutely time and energy intensive.

46
00:02:28,639 --> 00:02:31,640
Speaker 2: Researchers had to go through the papers one by one,

47
00:02:32,120 --> 00:02:35,359
Speaker 2: analyzing the data a step by step, and then spend

48
00:02:35,639 --> 00:02:39,799
Speaker 2: weeks in order to organize the information. But now today

49
00:02:40,000 --> 00:02:43,440
Speaker 2: AI can automate many tasks such as a writing a

50
00:02:43,520 --> 00:02:48,079
Speaker 2: literatariy reviewer, can summarize hundreds of papers such as research

51
00:02:48,160 --> 00:02:52,080
Speaker 2: a direction or even methodologies, and even can assist in

52
00:02:52,160 --> 00:02:57,040
Speaker 2: the coding or analyzing and ultimately reducing weeks of work

53
00:02:57,200 --> 00:03:01,239
Speaker 2: into the hours. So these allows of researchers to focus

54
00:03:01,280 --> 00:03:06,600
Speaker 2: more on insight, creativity, and innovation rather than repetitive tasks.

55
00:03:07,040 --> 00:03:10,480
Speaker 1: I get you now in terms of extrapulating data and

56
00:03:10,520 --> 00:03:12,760
Speaker 1: making sense of that data. I think that's what makes

57
00:03:12,760 --> 00:03:15,320
Speaker 1: it so impressive really is not only can it filter

58
00:03:15,440 --> 00:03:20,400
Speaker 1: through masses and masses of data incredibly fast, it can

59
00:03:20,440 --> 00:03:23,160
Speaker 1: then make sense of that and bring you conclusions. It

60
00:03:23,160 --> 00:03:27,320
Speaker 1: can bring you essentially exactly what you're looking for throughout

61
00:03:27,320 --> 00:03:29,520
Speaker 1: all these vast quantities of data, and it can do so,

62
00:03:29,600 --> 00:03:34,240
Speaker 1: of course at light speeds. This obviously ups on your productivity,

63
00:03:34,320 --> 00:03:38,040
Speaker 1: but does come with risks, doesn't it? Risks of accuracy,

64
00:03:38,360 --> 00:03:43,160
Speaker 1: risks of complete hallucinations that we know AI is prone to.

65
00:03:43,560 --> 00:03:46,200
Speaker 1: How do you know that? Surely? Is the job of

66
00:03:46,240 --> 00:03:48,480
Speaker 1: the researcher then is to ensure that that data that's

67
00:03:48,520 --> 00:03:51,840
Speaker 1: being extrapulated is still quality data.

68
00:03:52,120 --> 00:03:56,600
Speaker 2: Yes, absolutely, So AI also comes with some risk injur research.

69
00:03:56,680 --> 00:04:00,040
Speaker 2: So the main risk are all the real lives and

70
00:04:00,120 --> 00:04:05,040
Speaker 2: of critical evaluations. So AI can generate very convincing but

71
00:04:05,120 --> 00:04:07,720
Speaker 2: at the same time convincing output, but at the same

72
00:04:07,720 --> 00:04:11,120
Speaker 2: time there are not always accurate and reliable. So if

73
00:04:11,240 --> 00:04:15,720
Speaker 2: researchers accept the result without a questioning them, and it

74
00:04:15,760 --> 00:04:19,520
Speaker 2: can lead to the error or even flowed conclusion. And

75
00:04:19,560 --> 00:04:22,480
Speaker 2: there are also some some concerns about it, about a

76
00:04:22,560 --> 00:04:26,800
Speaker 2: bias in the data and also transparency of AI models.

77
00:04:26,600 --> 00:04:30,240
Speaker 2: That's why we share. We say that it is essential

78
00:04:30,320 --> 00:04:33,680
Speaker 2: to treat AI as a support tool, but rather not

79
00:04:33,760 --> 00:04:37,840
Speaker 2: to rely on the AI. So so researchers must validate

80
00:04:37,880 --> 00:04:42,520
Speaker 2: the result and make sure to maintain a strong scientific rigor.

81
00:04:43,520 --> 00:04:47,120
Speaker 1: I've heard some incredibly funny things that AI has come

82
00:04:47,200 --> 00:04:49,919
Speaker 1: up with. I'm a chat GBT user myself. Just the

83
00:04:49,960 --> 00:04:52,360
Speaker 1: other day I asked a question I don't remember once

84
00:04:52,400 --> 00:04:55,120
Speaker 1: and it gave me an answer, and its answer was say,

85
00:04:55,279 --> 00:04:59,200
Speaker 1: a full paragraph answer, but the second paragraph was written

86
00:04:59,279 --> 00:05:04,760
Speaker 1: in Greek for no particular reason at all. So it's imperfect,

87
00:05:04,839 --> 00:05:05,279
Speaker 1: isn't it.

88
00:05:06,800 --> 00:05:11,680
Speaker 2: Of course it is not. It's AI. AI can can significantly,

89
00:05:12,360 --> 00:05:16,039
Speaker 2: you know, affect productivity, but it is it is just

90
00:05:16,080 --> 00:05:19,159
Speaker 2: a tool to help you to become a better version

91
00:05:19,200 --> 00:05:22,280
Speaker 2: of yourself. In fact, it is a tool to augment

92
00:05:22,480 --> 00:05:27,960
Speaker 2: human intelligence and not not not to replace it. So

93
00:05:27,960 --> 00:05:30,479
Speaker 2: so it doesn't it doesn't replace the human being, and

94
00:05:30,520 --> 00:05:33,839
Speaker 2: it is not a perfect tool to completely rely on

95
00:05:33,920 --> 00:05:38,400
Speaker 2: it because still human judgment and critical thinking and creativity

96
00:05:38,600 --> 00:05:42,800
Speaker 2: remains essential. So so the task is like like a

97
00:05:42,880 --> 00:05:47,560
Speaker 2: summarize normal taxes or repetitive tests, such as summarizing a paper,

98
00:05:47,760 --> 00:05:51,440
Speaker 2: organizing the references. In terms of the research I'm talking

99
00:05:51,480 --> 00:05:54,960
Speaker 2: to you, it's exploring data sets or even extracting the

100
00:05:54,960 --> 00:05:57,320
Speaker 2: a taste that can can be done in a fraction

101
00:05:57,480 --> 00:06:01,280
Speaker 2: of the time for the research to to Let's this

102
00:06:01,480 --> 00:06:05,360
Speaker 2: allow researchers to focus on critical thinking and problem solving,

103
00:06:06,120 --> 00:06:10,000
Speaker 2: and at the end of the day, the real shift

104
00:06:10,120 --> 00:06:15,440
Speaker 2: is toward a human AI collaboration and rather than being

105
00:06:15,520 --> 00:06:20,080
Speaker 2: a replacement. But as I said, a researcher must validate

106
00:06:20,920 --> 00:06:24,760
Speaker 2: the day result and question whatever they received from AI.

107
00:06:25,120 --> 00:06:27,320
Speaker 2: You cannot really one hundreds person to AI.

108
00:06:27,600 --> 00:06:31,720
Speaker 1: No, absolutely not. And I think your your keywords their

109
00:06:31,760 --> 00:06:34,680
Speaker 1: critical thinking and human judgment. And that's why. And I

110
00:06:34,720 --> 00:06:36,800
Speaker 1: think I've answered the next question I have in front

111
00:06:36,839 --> 00:06:39,919
Speaker 1: of me here, Will AI replace research? Is absolutely not

112
00:06:40,120 --> 00:06:43,400
Speaker 1: right because AI does not have human judgment or critical thinking.

113
00:06:43,440 --> 00:06:46,360
Speaker 1: I want if you could indulge me in a bit

114
00:06:46,360 --> 00:06:48,440
Speaker 1: of a real word example. I know that you've very

115
00:06:48,440 --> 00:06:52,520
Speaker 1: recently completed your PhD. Congratulations by the way, and I

116
00:06:52,600 --> 00:06:55,960
Speaker 1: assume that AI played a huge role in the work

117
00:06:56,000 --> 00:06:58,240
Speaker 1: that you do actively every day, not just in the

118
00:06:58,279 --> 00:07:01,479
Speaker 1: workshop that you're hosting about AI, but but actively in

119
00:07:01,520 --> 00:07:03,760
Speaker 1: the work that you do at the university every day.

120
00:07:05,320 --> 00:07:08,000
Speaker 2: Yes, yes, I bury you all right, as I as

121
00:07:08,000 --> 00:07:10,520
Speaker 2: I mentioned, AI is not going to replace the human being,

122
00:07:10,560 --> 00:07:13,640
Speaker 2: but human being are going to be replaced by by

123
00:07:13,680 --> 00:07:16,920
Speaker 2: the human who embarrass the AI. So it is it

124
00:07:17,040 --> 00:07:19,560
Speaker 2: is like, as I said, the future is going to

125
00:07:19,560 --> 00:07:22,280
Speaker 2: be human AI collaboration. So if you are a human

126
00:07:22,360 --> 00:07:24,280
Speaker 2: human and you're you don't have a knowledge and you

127
00:07:24,320 --> 00:07:26,800
Speaker 2: are not going to adapt to AI, and you know

128
00:07:26,840 --> 00:07:28,960
Speaker 2: how to use the AI in your research, not only

129
00:07:28,960 --> 00:07:32,640
Speaker 2: in the research in many aspects. Then you will probably

130
00:07:32,720 --> 00:07:35,320
Speaker 2: replace in the in the near future with the human

131
00:07:35,360 --> 00:07:37,240
Speaker 2: who knows how to how to use the AI in

132
00:07:37,920 --> 00:07:40,960
Speaker 2: their freedom knowledge. In terms of the example, there are

133
00:07:41,000 --> 00:07:44,400
Speaker 2: plenty examples barriers. So at the moment I am in

134
00:07:44,440 --> 00:07:48,000
Speaker 2: the biomedical engineering research with it at the research units

135
00:07:48,120 --> 00:07:52,400
Speaker 2: as at U City Metach, so we are we are

136
00:07:52,400 --> 00:07:57,480
Speaker 2: focusing on the healthcare and patients. So in healthcare, AI

137
00:07:57,600 --> 00:08:02,240
Speaker 2: can be used in analyzing like a medical images, detect

138
00:08:02,280 --> 00:08:06,360
Speaker 2: patterns in the patient data, and even accelerate drug discovery.

139
00:08:06,720 --> 00:08:11,680
Speaker 2: It allows research and clinations to identify insights that would

140
00:08:11,720 --> 00:08:16,000
Speaker 2: be extremely difficult to detect to be detected by by

141
00:08:16,080 --> 00:08:20,560
Speaker 2: human being and alone. Because AI can detect hidden patterns,

142
00:08:20,720 --> 00:08:23,960
Speaker 2: complex pattern can be trained and find out the hidden

143
00:08:24,000 --> 00:08:28,520
Speaker 2: patterns that can be cannot be seen by and found

144
00:08:28,600 --> 00:08:33,079
Speaker 2: by a human being. So it enables more accurate, faster

145
00:08:33,320 --> 00:08:38,160
Speaker 2: and data driven decision making which ultimately improves patients outcome.

146
00:08:38,360 --> 00:08:41,800
Speaker 2: So I have been using AI in mechanical and materials engineering.

147
00:08:41,840 --> 00:08:45,120
Speaker 2: So AI can optimize the designe It can be a

148
00:08:45,200 --> 00:08:49,839
Speaker 2: multi objective optimizations in a mechanical engineer, or we can

149
00:08:49,880 --> 00:08:53,520
Speaker 2: predict the material material behavior by using the AI, so

150
00:08:53,600 --> 00:08:56,920
Speaker 2: depend on the task, we will use different algorithms and

151
00:08:57,160 --> 00:09:02,640
Speaker 2: architectures in order to come up with the accurate outcome.

152
00:09:03,160 --> 00:09:06,720
Speaker 2: A I also can can scan hundreds of papers in

153
00:09:06,800 --> 00:09:10,240
Speaker 2: terms of well in in our research in minutes to

154
00:09:10,400 --> 00:09:15,120
Speaker 2: identify the research gap and can help to structure literature,

155
00:09:15,200 --> 00:09:20,360
Speaker 2: review and event suggest methodology or high particles. What takes

156
00:09:20,400 --> 00:09:23,520
Speaker 2: a weeks and now can be done just in our aim.

157
00:09:23,760 --> 00:09:27,040
Speaker 2: Very in is no longer optional. Very it is not

158
00:09:27,080 --> 00:09:30,680
Speaker 2: only in not only in research, but even in industries

159
00:09:30,720 --> 00:09:34,640
Speaker 2: and many aspects. So it is becoming a core competency.

160
00:09:35,080 --> 00:09:39,280
Speaker 2: So researchers who don't or in general, people who don't adapt,

161
00:09:39,760 --> 00:09:43,280
Speaker 2: they risk falling behind. And while those those people and

162
00:09:43,320 --> 00:09:47,839
Speaker 2: researchers who embarrassed and leverage AI can dramatically increase their

163
00:09:47,880 --> 00:09:53,080
Speaker 2: productivity and can uncover the insights faster and estate competitive

164
00:09:53,120 --> 00:09:53,680
Speaker 2: gota value.

165
00:09:53,840 --> 00:09:54,080
Speaker 3: Yeah.

166
00:09:54,320 --> 00:09:58,000
Speaker 1: I think accepting and embracing AI at the moment is

167
00:09:58,160 --> 00:10:01,240
Speaker 1: like the acceptance and embracing of the Internet when we

168
00:10:01,320 --> 00:10:03,440
Speaker 1: got the Internet and a lot of people were like, no,

169
00:10:03,960 --> 00:10:06,839
Speaker 1: this is a phase, it will pass and now look

170
00:10:07,040 --> 00:10:08,679
Speaker 1: if you've just joined us. By the way, we're speaking

171
00:10:08,720 --> 00:10:12,160
Speaker 1: to doctor Samilagoliza. There's materials and mechanical engineer and an

172
00:10:12,200 --> 00:10:15,680
Speaker 1: AI specialist at the University of Cape Town. She's also

173
00:10:15,720 --> 00:10:18,520
Speaker 1: a postdoctoral research fellow. And we're talking AI and how

174
00:10:18,600 --> 00:10:21,920
Speaker 1: AI is changing the game for research and continues to

175
00:10:22,000 --> 00:10:25,360
Speaker 1: change most everything in this world now. So we're talking

176
00:10:25,400 --> 00:10:28,160
Speaker 1: about the skills that researchers will have to adopt in

177
00:10:28,320 --> 00:10:32,360
Speaker 1: order to continue on with their research by using these tools. Skills,

178
00:10:32,400 --> 00:10:35,880
Speaker 1: as you said, like using perhaps a different method of

179
00:10:35,880 --> 00:10:40,880
Speaker 1: critical thinking and human judgment. We said, AI is perfectly imperfect,

180
00:10:41,160 --> 00:10:44,679
Speaker 1: but it can extrapulate through vast quantities of data and

181
00:10:44,720 --> 00:10:48,760
Speaker 1: give you calculated responses. So I assume that the new skills,

182
00:10:48,800 --> 00:10:51,079
Speaker 1: and I hope you'll elaborate on this, the new skill

183
00:10:51,120 --> 00:10:54,240
Speaker 1: is really just being able to how can I say

184
00:10:54,280 --> 00:10:57,200
Speaker 1: this is to rain in on the AI and to

185
00:10:57,440 --> 00:11:00,439
Speaker 1: ensure that the information that you're getting is in fact accurate.

186
00:11:00,679 --> 00:11:03,520
Speaker 1: What about referencing? By the way, I have a voice

187
00:11:03,600 --> 00:11:04,800
Speaker 1: note that I'd like to play for us in a

188
00:11:04,880 --> 00:11:07,640
Speaker 1: little while, but I'll get to the voice note. When

189
00:11:07,679 --> 00:11:11,160
Speaker 1: it comes to research, we always have to reference our work.

190
00:11:11,240 --> 00:11:13,200
Speaker 1: How do we do that when we're using AI to

191
00:11:13,240 --> 00:11:16,199
Speaker 1: go through these vast quantities of data. Is it going

192
00:11:16,240 --> 00:11:18,360
Speaker 1: to give you accurate references for where it found all

193
00:11:18,360 --> 00:11:19,040
Speaker 1: that information?

194
00:11:21,000 --> 00:11:24,000
Speaker 2: So, Bari, in terms of the skills, you're right, the

195
00:11:24,200 --> 00:11:27,800
Speaker 2: sculls is going to be a shifted. So as I said,

196
00:11:28,040 --> 00:11:32,320
Speaker 2: there is some some replacement in some repetitive tasks that's

197
00:11:32,360 --> 00:11:35,040
Speaker 2: going to be done by AI, but we still have

198
00:11:35,200 --> 00:11:38,400
Speaker 2: a new skills that era a bring along with somes

199
00:11:38,800 --> 00:11:42,440
Speaker 2: with itself. So as a researchers, do they need a

200
00:11:42,480 --> 00:11:47,480
Speaker 2: combination of a different skill like a technical and cognitive skills? First,

201
00:11:47,640 --> 00:11:52,280
Speaker 2: AI literacy is AI literacy understanding how to use the

202
00:11:52,320 --> 00:11:57,120
Speaker 2: AI tools effectively and responsibly along with the strong critical

203
00:11:57,280 --> 00:12:00,400
Speaker 2: thinking as you mentioned, and data into approach. It is

204
00:12:00,440 --> 00:12:04,880
Speaker 2: skilled because AI output still need to be evaluated carefully.

205
00:12:05,240 --> 00:12:08,800
Speaker 1: Now I've got a question. I'm so sorry I've interrupted you, Samita.

206
00:12:08,880 --> 00:12:10,440
Speaker 1: We've got a question here on the one step plan.

207
00:12:10,520 --> 00:12:12,440
Speaker 1: I'd love to pass your way. It's not really a question,

208
00:12:12,520 --> 00:12:15,080
Speaker 1: it's a statement, and it says listening to this AI

209
00:12:15,200 --> 00:12:18,760
Speaker 1: segment and how students can blatantly use AI and actually

210
00:12:18,880 --> 00:12:22,240
Speaker 1: expect to pass. Tell me how how do we get

211
00:12:22,280 --> 00:12:24,800
Speaker 1: around this. I've now heard more than one account where

212
00:12:24,880 --> 00:12:28,560
Speaker 1: universities and institutions of learning discourage the use of AI

213
00:12:28,720 --> 00:12:31,280
Speaker 1: or will actually you know, fail you based on the

214
00:12:31,320 --> 00:12:34,440
Speaker 1: fact of using AI. Why is that? Why is there

215
00:12:34,480 --> 00:12:37,000
Speaker 1: not yet a general acceptance of AI as.

216
00:12:36,800 --> 00:12:41,760
Speaker 2: A tool, so so barring I think is that we

217
00:12:41,840 --> 00:12:44,959
Speaker 2: need to introduce AI and we should accept the AI

218
00:12:45,000 --> 00:12:48,360
Speaker 2: in the academic and introduce it to to the students

219
00:12:48,200 --> 00:12:53,240
Speaker 2: as effective tool. A student shouldn't use AI blindly, you know.

220
00:12:53,360 --> 00:12:56,040
Speaker 2: So I think if you just rely one hundred percent

221
00:12:56,080 --> 00:12:58,520
Speaker 2: of the AI and just copy and paste whatever you

222
00:12:58,600 --> 00:13:02,720
Speaker 2: received from AI, that can inaccurate. It can the result

223
00:13:02,800 --> 00:13:05,959
Speaker 2: that output must be questions and must be validated through

224
00:13:06,120 --> 00:13:10,440
Speaker 2: the literashire review, through the research, through the study. So

225
00:13:10,440 --> 00:13:14,040
Speaker 2: so we have at the moment in any research and

226
00:13:14,080 --> 00:13:16,640
Speaker 2: in general, from the beginning to end to the publication,

227
00:13:16,760 --> 00:13:20,319
Speaker 2: we have I think one million AI tools available around

228
00:13:20,360 --> 00:13:22,880
Speaker 2: the world that can be used in the academic. But

229
00:13:22,920 --> 00:13:25,360
Speaker 2: it doesn't mean that we need to introduce one million

230
00:13:25,400 --> 00:13:27,880
Speaker 2: tools today to the students. The thing is act when

231
00:13:27,920 --> 00:13:30,880
Speaker 2: we talk about the AI in academic, everyone thinking about

232
00:13:30,880 --> 00:13:34,679
Speaker 2: the chadgivity. Every term one I think about think about

233
00:13:34,679 --> 00:13:38,880
Speaker 2: the cloud or peroxiality. So it is these tools are

234
00:13:39,080 --> 00:13:41,880
Speaker 2: very general AI, so I can so, so whoever to

235
00:13:42,080 --> 00:13:44,920
Speaker 2: talk to me about the chatibilty, I usually respond like

236
00:13:45,200 --> 00:13:49,000
Speaker 2: A and chagibility is the Google is advanced level of

237
00:13:49,040 --> 00:13:51,960
Speaker 2: a Google. It's advanced level of a research engine. So

238
00:13:52,120 --> 00:13:55,600
Speaker 2: if you want to reach out to something quickly, chatchivity

239
00:13:55,679 --> 00:13:57,920
Speaker 2: is the best is the best AI tools in order

240
00:13:57,960 --> 00:13:59,760
Speaker 2: to search for But it is not something that you

241
00:13:59,800 --> 00:14:01,959
Speaker 2: can real life or the research you know. It can

242
00:14:02,000 --> 00:14:05,280
Speaker 2: be completely inaccurate. So if you go through the different

243
00:14:05,320 --> 00:14:07,840
Speaker 2: examples we can find it out. But if you know

244
00:14:07,920 --> 00:14:11,599
Speaker 2: the effective ALE tools in your research or even in

245
00:14:12,520 --> 00:14:15,559
Speaker 2: your industries, if you find out which one is effective.

246
00:14:15,600 --> 00:14:19,280
Speaker 2: So effective means that we have we have specific al

247
00:14:19,360 --> 00:14:23,200
Speaker 2: tools for brainstorming for the idea generations, which is the

248
00:14:23,200 --> 00:14:26,600
Speaker 2: beginning of a research. We have effective ALE tools very

249
00:14:26,720 --> 00:14:29,840
Speaker 2: specific that can analyze and extract the data. We have

250
00:14:29,880 --> 00:14:34,280
Speaker 2: the AI tools vary that that can compare different papers

251
00:14:34,320 --> 00:14:37,480
Speaker 2: and extract the data from the papers which are relevant

252
00:14:37,560 --> 00:14:41,320
Speaker 2: to each other. And it can connect the data from

253
00:14:41,360 --> 00:14:44,000
Speaker 2: the different papers. Then it can show you as a

254
00:14:44,040 --> 00:14:46,920
Speaker 2: bar chart or as a graph or a different chart

255
00:14:47,040 --> 00:14:50,760
Speaker 2: to show you how different results look like. If it

256
00:14:50,800 --> 00:14:54,840
Speaker 2: can help you to find out your answer reading these results.

257
00:14:55,000 --> 00:14:58,200
Speaker 2: And also we can we can have we have effective

258
00:14:58,280 --> 00:15:00,880
Speaker 2: tools that can give you a methodol how to you

259
00:15:01,080 --> 00:15:04,360
Speaker 2: how to do your your research or if you have

260
00:15:04,840 --> 00:15:07,920
Speaker 2: a doubt about your high partestlers, you can ask AI

261
00:15:08,080 --> 00:15:12,160
Speaker 2: and you can ask help and go to different AI

262
00:15:12,200 --> 00:15:16,400
Speaker 2: tools which is designed specific for a high partesisers. So

263
00:15:16,640 --> 00:15:21,520
Speaker 2: and ultimately structure your your paper or literature review that

264
00:15:21,600 --> 00:15:22,720
Speaker 2: come up with a publication.

265
00:15:23,040 --> 00:15:25,760
Speaker 1: Yeah, what I'm hearing there is that not all AIS

266
00:15:25,800 --> 00:15:27,600
Speaker 1: are made equal. And before I get to my question,

267
00:15:27,760 --> 00:15:30,600
Speaker 1: just a reminder if you're also welcome to ask Samita

268
00:15:30,600 --> 00:15:33,120
Speaker 1: Goliza there who I'm speaking to now a question that

269
00:15:33,480 --> 00:15:36,680
Speaker 1: any question that you'd like in relation to AI. You

270
00:15:36,720 --> 00:15:38,320
Speaker 1: can give me a call if you'd like. It's O

271
00:15:38,440 --> 00:15:41,080
Speaker 1: two one four four six O five six seven, or

272
00:15:41,120 --> 00:15:43,080
Speaker 1: you can once pp O seven two five six seven

273
00:15:43,160 --> 00:15:45,240
Speaker 1: one five six seven. So you know what I'm hearing

274
00:15:45,280 --> 00:15:48,360
Speaker 1: there is that not all AIS are created equal. And

275
00:15:48,400 --> 00:15:52,520
Speaker 1: you would probably use one AI. How can I say

276
00:15:52,560 --> 00:15:55,520
Speaker 1: you you would use one application for one function and

277
00:15:55,560 --> 00:15:58,640
Speaker 1: then switch to a different application for a different function.

278
00:16:00,280 --> 00:16:02,840
Speaker 2: Not one application, it can be one hundred applications, but

279
00:16:03,000 --> 00:16:06,800
Speaker 2: how you should find out what is your task? What

280
00:16:06,840 --> 00:16:09,400
Speaker 2: do you need, what is your output? What is your expectation?

281
00:16:09,560 --> 00:16:13,360
Speaker 2: Then you define and you decide which kind of AI

282
00:16:13,520 --> 00:16:16,720
Speaker 2: tools is more effective in your task. So it's not

283
00:16:16,800 --> 00:16:18,640
Speaker 2: one application, it's is one hundred and it can be

284
00:16:18,640 --> 00:16:21,400
Speaker 2: a thousands. But you pick up on yourself based on

285
00:16:21,440 --> 00:16:24,160
Speaker 2: your task and the output that you expect from.

286
00:16:24,400 --> 00:16:26,800
Speaker 1: But I mean, so for someone like me who is

287
00:16:26,840 --> 00:16:30,320
Speaker 1: not a researcher, and I use Chatgypt and I maybe

288
00:16:30,360 --> 00:16:33,320
Speaker 1: know the names of like two other ais like Grock

289
00:16:33,520 --> 00:16:36,640
Speaker 1: or Claude, what are the kinds of tools that researchers

290
00:16:36,680 --> 00:16:39,760
Speaker 1: are using at UCT? Are they? Are they apps also

291
00:16:39,800 --> 00:16:42,840
Speaker 1: publicly available to people like me and ob and shay

292
00:16:43,280 --> 00:16:48,320
Speaker 1: or are these specifically developed and owned by institutions? Where

293
00:16:48,360 --> 00:16:51,320
Speaker 1: are they.

294
00:16:50,240 --> 00:16:54,560
Speaker 2: Now at the moment, i usity is not implementing a

295
00:16:54,920 --> 00:16:59,320
Speaker 2: Chatgypt or growth or a cloud specifically for students. No,

296
00:16:59,480 --> 00:17:02,960
Speaker 2: there are as a general AI tools that available worldwide

297
00:17:02,960 --> 00:17:06,040
Speaker 2: for anyone else that based on based on they need.

298
00:17:07,280 --> 00:17:11,240
Speaker 2: But it is like it is like a general general

299
00:17:11,280 --> 00:17:14,920
Speaker 2: AI tool that can that students can can use freely.

300
00:17:15,280 --> 00:17:17,800
Speaker 2: But if you want to go deeper, if if if

301
00:17:17,840 --> 00:17:20,600
Speaker 2: you check it, check those those tools. If you want

302
00:17:20,600 --> 00:17:23,480
Speaker 2: to go to deeper, you have to pay. If you pay,

303
00:17:23,600 --> 00:17:27,720
Speaker 2: then maybe if they're more accurate or more deep deep

304
00:17:27,800 --> 00:17:30,719
Speaker 2: results or deep analyzers from the tools. But it is

305
00:17:30,720 --> 00:17:33,480
Speaker 2: not connected to the to the U city platform yet.

306
00:17:33,680 --> 00:17:36,360
Speaker 2: And I don't think so that any education is going

307
00:17:36,400 --> 00:17:41,360
Speaker 2: to impediment or connect those general tools to their website

308
00:17:41,480 --> 00:17:46,400
Speaker 2: or to their platform. But students at ucity are allowed

309
00:17:46,440 --> 00:17:52,119
Speaker 2: to use AI into their research effectively and more responsibily,

310
00:17:52,320 --> 00:17:55,520
Speaker 2: so it is not they are not allowed to use

311
00:17:55,560 --> 00:17:59,480
Speaker 2: it blindly. And also, and I think it's a US

312
00:17:59,480 --> 00:18:03,439
Speaker 2: city A lots students to use AI buys. Students are

313
00:18:03,480 --> 00:18:07,679
Speaker 2: responsible to declaric that they have been using the AI

314
00:18:07,840 --> 00:18:08,760
Speaker 2: for their research.

315
00:18:09,200 --> 00:18:12,560
Speaker 1: Right as we got a voice note in earlier from

316
00:18:12,640 --> 00:18:15,520
Speaker 1: ras Metcalfe, and I actually see that we got another one,

317
00:18:15,520 --> 00:18:17,119
Speaker 1: and I'm sure Shaye will tell us what's in it

318
00:18:17,160 --> 00:18:20,560
Speaker 1: in just a second. Where some students aren't being penalized

319
00:18:20,560 --> 00:18:22,560
Speaker 1: for the use of AI or for the suspicion of

320
00:18:22,680 --> 00:18:25,560
Speaker 1: using AI, I can tell you this much. I am

321
00:18:26,240 --> 00:18:28,879
Speaker 1: so reliant on my chat GPT. I'm attached to my

322
00:18:29,000 --> 00:18:31,600
Speaker 1: Chat GPT, I do pay for the premium version of it,

323
00:18:31,680 --> 00:18:33,800
Speaker 1: and I use it every single day of my life.

324
00:18:33,880 --> 00:18:36,159
Speaker 1: It's become such an essential to me. And I know

325
00:18:36,200 --> 00:18:38,760
Speaker 1: I'm not the only one at all, and I'm not

326
00:18:38,840 --> 00:18:41,920
Speaker 1: even a researcher, right I work here. When it comes

327
00:18:42,000 --> 00:18:43,479
Speaker 1: to the work that you do and some of your

328
00:18:43,480 --> 00:18:46,080
Speaker 1: colleagues do, I can only imagine it's roughly the same.

329
00:18:46,880 --> 00:18:49,359
Speaker 1: Moving slightly on in this conversation, I'd like to know

330
00:18:49,400 --> 00:18:52,000
Speaker 1: what this will mean for the future of science, meaning,

331
00:18:52,040 --> 00:18:54,640
Speaker 1: as you said, what used to be an incredibly manual,

332
00:18:54,760 --> 00:18:58,400
Speaker 1: incredibly tedious task of research having to go through vast

333
00:18:58,480 --> 00:19:01,640
Speaker 1: quantities of data now happened really fast. And that probably

334
00:19:01,680 --> 00:19:05,359
Speaker 1: also means that science and research is going to move

335
00:19:05,440 --> 00:19:08,200
Speaker 1: forward at the pace that we probably can't imagine just yet.

336
00:19:08,480 --> 00:19:11,880
Speaker 2: Oh so, Barie, it refers when we call we talk

337
00:19:11,960 --> 00:19:14,760
Speaker 2: about the future of science and research, it means that

338
00:19:15,320 --> 00:19:20,400
Speaker 2: we why be implementing AI. It means that like faster discoveries,

339
00:19:20,880 --> 00:19:26,480
Speaker 2: more interdisciplinary research, and greater accessibility, AI can democratize research

340
00:19:26,680 --> 00:19:31,520
Speaker 2: by giving more people power, powerful tools to contribute to

341
00:19:31,560 --> 00:19:35,920
Speaker 2: the to the science and and chat to pet And

342
00:19:36,160 --> 00:19:38,960
Speaker 2: as you mentioned, it is it is something like it

343
00:19:39,080 --> 00:19:41,680
Speaker 2: is when when you set to to someone that don't

344
00:19:41,800 --> 00:19:44,159
Speaker 2: use AI, it is like to say to someone that

345
00:19:44,240 --> 00:19:49,000
Speaker 2: don't use Internet. So so AI nowadays is like Internet

346
00:19:49,040 --> 00:19:52,119
Speaker 2: a few years ago. You know, so so it is.

347
00:19:52,920 --> 00:19:57,000
Speaker 2: But the thing is we should know why we are

348
00:19:57,080 --> 00:20:00,400
Speaker 2: using a I and what is the purpose of using

349
00:20:01,119 --> 00:20:04,480
Speaker 2: a AI in research in the science, and how it's

350
00:20:04,520 --> 00:20:10,400
Speaker 2: going to affect our productivity in many aspects.

351
00:20:10,440 --> 00:20:13,040
Speaker 1: Interesting interesting stuff, A lot of food for thought. Then,

352
00:20:13,359 --> 00:20:15,320
Speaker 1: while I was preparing for this discussion, I was I

353
00:20:15,359 --> 00:20:17,520
Speaker 1: was giving it a lot of thought and reading up

354
00:20:17,520 --> 00:20:19,560
Speaker 1: a lot about it. And as you say, it is

355
00:20:19,560 --> 00:20:22,320
Speaker 1: the new Internet, and it's rapidly, rapidly changing, and it's

356
00:20:22,359 --> 00:20:25,400
Speaker 1: becoming better and better every single day. I don't think

357
00:20:25,400 --> 00:20:28,600
Speaker 1: that we even have an idea of just how how

358
00:20:28,680 --> 00:20:31,239
Speaker 1: much it's going to change the game for research and

359
00:20:31,280 --> 00:20:34,640
Speaker 1: for science within various different fields. I wonder, Samita, while

360
00:20:34,680 --> 00:20:36,240
Speaker 1: we've got you on the line and you are listening

361
00:20:36,240 --> 00:20:38,400
Speaker 1: to the voice of doctor samiit a Goliza there. She's

362
00:20:38,400 --> 00:20:42,320
Speaker 1: in materials and mechanical engineer and AI specialist with uct Obi.

363
00:20:42,359 --> 00:20:44,080
Speaker 1: Do we have that voice note from rass that we

364
00:20:44,119 --> 00:20:46,040
Speaker 1: played earlier. I'd actually like to play it again, if

365
00:20:46,080 --> 00:20:48,919
Speaker 1: that's okay for Samita to hear. When you have a

366
00:20:48,960 --> 00:20:50,320
Speaker 1: listen at this Hi.

367
00:20:51,359 --> 00:20:55,720
Speaker 3: Every time I hear this mention of artificial intelligence, I

368
00:20:55,800 --> 00:21:01,359
Speaker 3: get riled because I am an elderly student at the

369
00:21:01,480 --> 00:21:05,800
Speaker 3: University of South Africa. I have been accused of cheating

370
00:21:06,000 --> 00:21:11,080
Speaker 3: and using AI for answering some of my questions which

371
00:21:11,200 --> 00:21:14,480
Speaker 3: I submitted last year, and I was given two percent.

372
00:21:15,119 --> 00:21:18,359
Speaker 3: I was told that I could appeal this. I did,

373
00:21:18,520 --> 00:21:22,960
Speaker 3: and have heard absolutely nothing from the university. I'd really

374
00:21:22,960 --> 00:21:26,920
Speaker 3: be grateful if you could give my number to any

375
00:21:27,359 --> 00:21:32,080
Speaker 3: university student who have suffered similarly. I've basically ended my

376
00:21:32,400 --> 00:21:38,320
Speaker 3: university career because of this, and I've got absolutely no

377
00:21:38,520 --> 00:21:42,640
Speaker 3: answer from the psychology department at UNISSA. And I will

378
00:21:42,720 --> 00:21:46,199
Speaker 3: keep on because I feel as an elderly student, I

379
00:21:46,320 --> 00:21:52,000
Speaker 3: owe it to my fellow younger students to object when

380
00:21:52,280 --> 00:21:57,200
Speaker 3: we are unfairly treated in this way. It's rush from Thornton,

381
00:21:57,440 --> 00:21:58,440
Speaker 3: Thanks so much.

382
00:21:58,920 --> 00:22:00,480
Speaker 1: And so you know, I would just like and that

383
00:22:00,600 --> 00:22:02,879
Speaker 1: Ros then later clarified on Takets that he did not

384
00:22:03,160 --> 00:22:06,520
Speaker 1: use AI in his work, and so I think that,

385
00:22:06,960 --> 00:22:09,840
Speaker 1: you know not. I think we're clearly seeing that there's

386
00:22:09,920 --> 00:22:12,719
Speaker 1: still a little bit of what do we call it,

387
00:22:12,840 --> 00:22:14,760
Speaker 1: There's I don't want to call it pushback.

388
00:22:14,920 --> 00:22:19,480
Speaker 2: That's all, yeah, Barry, you know what. Sorry sorry for interruption,

389
00:22:19,680 --> 00:22:21,800
Speaker 2: but something came to my mind that we should we

390
00:22:21,800 --> 00:22:24,959
Speaker 2: should be very careful about how we use the AI ethically.

391
00:22:25,040 --> 00:22:28,080
Speaker 2: You know, ethics in AI and ethics in research are

392
00:22:28,160 --> 00:22:31,240
Speaker 2: very important. So you can carry on your question and

393
00:22:31,240 --> 00:22:32,560
Speaker 2: then I can answer you.

394
00:22:32,600 --> 00:22:34,440
Speaker 1: No, I guess that was the question to the question?

395
00:22:34,520 --> 00:22:36,960
Speaker 1: Really is your response to to Ross's message and what

396
00:22:37,080 --> 00:22:37,679
Speaker 1: you think of that?

397
00:22:39,400 --> 00:22:43,400
Speaker 2: So, Barry, I don't believe so. And let's say that

398
00:22:43,560 --> 00:22:47,600
Speaker 2: there are many universities that are very asterical about using

399
00:22:47,640 --> 00:22:51,800
Speaker 2: AI in research because they think if people think that

400
00:22:51,880 --> 00:22:54,760
Speaker 2: AI can be one hundred percent accurate and reliable, but

401
00:22:54,840 --> 00:22:57,959
Speaker 2: which is not. You know, if you asked a question

402
00:22:58,400 --> 00:23:01,679
Speaker 2: from AI and just copy paste, we have AI detect

403
00:23:01,760 --> 00:23:05,120
Speaker 2: tools that can be used and find out that your

404
00:23:05,160 --> 00:23:10,280
Speaker 2: answer is true. AI is generated by AI, so that

405
00:23:10,280 --> 00:23:14,199
Speaker 2: that tools can can detect the general content that is

406
00:23:14,560 --> 00:23:18,840
Speaker 2: generated by AI. So I think it's where AI is

407
00:23:18,880 --> 00:23:22,240
Speaker 2: here in order to help you. But how to how

408
00:23:22,280 --> 00:23:27,320
Speaker 2: to ask the right question and frame a meaningful problems

409
00:23:27,359 --> 00:23:31,399
Speaker 2: and guide the direction of the inquiry all is important,

410
00:23:31,560 --> 00:23:33,359
Speaker 2: and but at the same time we have to be

411
00:23:33,600 --> 00:23:38,600
Speaker 2: very responsible in order to evalidate the results to maintain

412
00:23:38,760 --> 00:23:43,280
Speaker 2: our scientific rigord. So we shouldn't accept the results as

413
00:23:43,320 --> 00:23:45,359
Speaker 2: it is. We have to go through it and we

414
00:23:45,560 --> 00:23:49,560
Speaker 2: we we should make up so I can say aparthhising

415
00:23:50,240 --> 00:23:52,600
Speaker 2: so to the answer if the answer is one hundred

416
00:23:52,640 --> 00:23:56,320
Speaker 2: person accurate, But we have to go through different papers

417
00:23:56,320 --> 00:23:59,080
Speaker 2: and different research and articles in order to find out

418
00:23:59,160 --> 00:24:02,560
Speaker 2: if the answer is accurate or not. But even if

419
00:24:02,560 --> 00:24:06,120
Speaker 2: the answer is accurate, because we have AI detect tools

420
00:24:06,400 --> 00:24:11,119
Speaker 2: that can detect those contents that you're generated by AI,

421
00:24:11,240 --> 00:24:16,959
Speaker 2: we have to change the whatever very received from AI tools,

422
00:24:16,960 --> 00:24:19,119
Speaker 2: we have to go through the different path raising and

423
00:24:19,200 --> 00:24:23,800
Speaker 2: make it more scientific because I believe that AI give

424
00:24:23,880 --> 00:24:27,480
Speaker 2: you very general answer rather than scientific. So if we

425
00:24:27,560 --> 00:24:30,960
Speaker 2: are part of a scientific field, we have to make

426
00:24:31,040 --> 00:24:36,159
Speaker 2: sure that that asser sounds academic and scientific rather than general.

427
00:24:36,440 --> 00:24:39,800
Speaker 1: Yeah, Doctor Samitra Goliza there, thank you very much for

428
00:24:39,840 --> 00:24:42,800
Speaker 1: your times. She's the materials and mechanical engineer and AI

429
00:24:42,840 --> 00:24:46,040
Speaker 1: specialist with the University of Cape Town and she's hosting

430
00:24:46,040 --> 00:24:49,199
Speaker 1: a workshop it's called the AI Revolution in Research Tools

431
00:24:49,240 --> 00:24:51,760
Speaker 1: for the Modern Scholar. That workshops on the twenty first

432
00:24:51,760 --> 00:24:55,120
Speaker 1: of April. Letter U se tea med Tech. I think

433
00:24:55,160 --> 00:24:56,920
Speaker 1: I should go and check it out myself.