1
00:00:00,160 --> 00:00:02,880
Speaker 1: Now food Stuff Speaking of crime, food Stuffs reckon that

2
00:00:02,880 --> 00:00:05,400
Speaker 1: their facial recognition trial has been a success. Hey, they've

3
00:00:05,400 --> 00:00:07,240
Speaker 1: been running this thing for six months, been keeping an

4
00:00:07,240 --> 00:00:09,920
Speaker 1: eye on it, running it in twenty five North Island stores.

5
00:00:10,240 --> 00:00:12,960
Speaker 1: They reckon, harm was reduced by sixteen percent and it

6
00:00:13,000 --> 00:00:16,400
Speaker 1: prevented more than one hundred serious install incidents, including assault.

7
00:00:16,600 --> 00:00:18,919
Speaker 1: Now the General Council for Food Stuffs North Island is

8
00:00:19,000 --> 00:00:22,120
Speaker 1: Julian Benefield had Julian Hi, Heather, is this about what

9
00:00:22,160 --> 00:00:22,840
Speaker 1: you were expecting?

10
00:00:24,600 --> 00:00:28,360
Speaker 2: Well, it was ultimately a trial to work out whether

11
00:00:29,080 --> 00:00:33,040
Speaker 2: facial recognition technology could help keep our people safer by

12
00:00:33,080 --> 00:00:35,640
Speaker 2: preventing repeat offendage from doing more harm. We've seen huge

13
00:00:35,640 --> 00:00:38,080
Speaker 2: increases in retail crime and we felt we needed to

14
00:00:38,080 --> 00:00:40,280
Speaker 2: do more to keep our people safe. It was a

15
00:00:40,320 --> 00:00:42,640
Speaker 2: genuine trial where we wanted to learn whether we could

16
00:00:42,720 --> 00:00:45,320
Speaker 2: use the technology safely to do so. We now have

17
00:00:45,440 --> 00:00:48,400
Speaker 2: the final report from the Child's Independent evaluator who has

18
00:00:48,400 --> 00:00:50,920
Speaker 2: found that facial recognition has reduced serious harm and the

19
00:00:50,960 --> 00:00:53,840
Speaker 2: child stores by needstimate of sixteen percent and has strong

20
00:00:54,240 --> 00:00:56,600
Speaker 2: public support justifying the use of the technology.

21
00:00:57,080 --> 00:00:59,480
Speaker 1: So with the nine who were misidentified, well all of

22
00:00:59,520 --> 00:01:02,200
Speaker 1: them trest No.

23
00:01:02,200 --> 00:01:05,000
Speaker 2: So out of firstly, you know, we didn't want any

24
00:01:05,280 --> 00:01:10,600
Speaker 2: misidentifications through this process, but unfortunately there were nine situations

25
00:01:10,720 --> 00:01:14,280
Speaker 2: where there was a misidentification. Out of the nine, there

26
00:01:14,280 --> 00:01:17,440
Speaker 2: were only two customers who are asked to leave the stores.

27
00:01:17,959 --> 00:01:21,120
Speaker 2: In all of those situations, we owned our mistake and

28
00:01:21,200 --> 00:01:24,520
Speaker 2: acknowledged our mistake and apologist drives as soon as we could.

29
00:01:24,920 --> 00:01:27,360
Speaker 2: And in each of those situations, it was a human era.

30
00:01:27,480 --> 00:01:29,920
Speaker 2: It was a manual failure and not a failure of

31
00:01:29,920 --> 00:01:33,759
Speaker 2: the systems. We've apologized to the customers, and we've looked

32
00:01:33,800 --> 00:01:36,480
Speaker 2: very hard the learnings of those situations and implemented training

33
00:01:36,720 --> 00:01:38,280
Speaker 2: and improved procedures as a result.

34
00:01:38,400 --> 00:01:40,080
Speaker 1: See one of them, the one that's based on rude

35
00:01:40,120 --> 00:01:41,959
Speaker 1: through is going to fight you guys even further. Ah,

36
00:01:42,120 --> 00:01:42,800
Speaker 1: did you see that.

37
00:01:44,040 --> 00:01:46,200
Speaker 2: Yeah, it's probably not something I can comment on at

38
00:01:46,200 --> 00:01:48,560
Speaker 2: this point in time, but ultimately we know it was

39
00:01:48,560 --> 00:01:52,640
Speaker 2: an unfortunate situation. As has been widely reported previously. We've

40
00:01:52,640 --> 00:01:56,000
Speaker 2: acknowledged our mistake and apologized and thought very deeply about

41
00:01:56,000 --> 00:01:59,360
Speaker 2: the situation and implemented and let's.

42
00:01:59,200 --> 00:02:02,440
Speaker 1: Be honest about it. Let's be honest about it. If

43
00:02:02,480 --> 00:02:04,800
Speaker 1: you're going to ask a couple of people to leave

44
00:02:04,800 --> 00:02:07,160
Speaker 1: who shouldn't leave. It's probably worth it to stop one

45
00:02:07,200 --> 00:02:10,040
Speaker 1: hundred assaults or one hundred incidents, Isn't it like you're

46
00:02:10,040 --> 00:02:11,360
Speaker 1: going to have some This is going to happen.

47
00:02:12,520 --> 00:02:14,320
Speaker 2: Yeah, And look, this was a genuine trial that we

48
00:02:14,400 --> 00:02:17,040
Speaker 2: knew we would learn from. We didn't want any misidentifications,

49
00:02:17,040 --> 00:02:19,880
Speaker 2: but to give it context, that was you know, nine

50
00:02:19,880 --> 00:02:23,240
Speaker 2: situations out of seventeen hundred and forty two alerts, right,

51
00:02:23,320 --> 00:02:25,000
Speaker 2: so we got it right over ninety eight percent at

52
00:02:25,000 --> 00:02:26,440
Speaker 2: the time. But we have learned from it and we

53
00:02:26,480 --> 00:02:26,959
Speaker 2: will improve.

54
00:02:27,040 --> 00:02:29,120
Speaker 1: Yeah, what happens next? So are you guys do you

55
00:02:29,120 --> 00:02:30,760
Speaker 1: guys get to make the final call on whether you

56
00:02:30,800 --> 00:02:32,480
Speaker 1: can actually roll this out across the country or is

57
00:02:32,480 --> 00:02:33,640
Speaker 1: that with the Privacy Commissioner.

58
00:02:34,280 --> 00:02:36,120
Speaker 2: Yeah, So a couple of months ago, at the end

59
00:02:36,120 --> 00:02:39,280
Speaker 2: of the trial, we decided to continue using FASH recognition

60
00:02:39,320 --> 00:02:42,000
Speaker 2: technology in the twenty five trial stores under the same

61
00:02:42,080 --> 00:02:44,880
Speaker 2: privacy and security protocols, and that was based on the

62
00:02:44,880 --> 00:02:47,880
Speaker 2: preliminary findings of independent evaluator at that point in time.

63
00:02:48,200 --> 00:02:52,720
Speaker 2: At this stage, we're now waiting for the OPCs to

64
00:02:52,760 --> 00:02:56,120
Speaker 2: publish their public inquiry into our trial. We want to

65
00:02:56,120 --> 00:02:57,959
Speaker 2: hear what they have to say, so we will review

66
00:02:57,960 --> 00:03:00,560
Speaker 2: what they have to say about their own cryer inquiry

67
00:03:00,600 --> 00:03:03,160
Speaker 2: before deciding on any further use of the technology. But

68
00:03:03,200 --> 00:03:05,079
Speaker 2: today we just wanted to be transparent or where we

69
00:03:05,160 --> 00:03:06,639
Speaker 2: landed with our final evaluation.

70
00:03:06,880 --> 00:03:08,320
Speaker 1: So who gets to make the call as to whether

71
00:03:08,320 --> 00:03:10,760
Speaker 1: you roll it out? You or the Privacy Commissioner.

72
00:03:11,760 --> 00:03:14,600
Speaker 2: Well, as a business, we have felt that, you know,

73
00:03:14,639 --> 00:03:16,560
Speaker 2: that the harm is so bad that we need to

74
00:03:16,560 --> 00:03:18,400
Speaker 2: do more about it. So we as a business made

75
00:03:18,400 --> 00:03:20,560
Speaker 2: a decision to continue using So.

76
00:03:20,480 --> 00:03:22,640
Speaker 1: You guys get the final say, you can decide if

77
00:03:22,639 --> 00:03:23,360
Speaker 1: you want to do it or not.

78
00:03:24,600 --> 00:03:26,360
Speaker 2: Well, you know, we've decided to use it in the

79
00:03:26,360 --> 00:03:28,799
Speaker 2: twenty five trial stores, but we do respect the role

80
00:03:28,840 --> 00:03:32,280
Speaker 2: of the OPC. We've worked constructively both before the trial

81
00:03:32,320 --> 00:03:34,520
Speaker 2: and then through the trial as they've done their inquiry

82
00:03:34,520 --> 00:03:37,720
Speaker 2: and we've done store visits and had you know, constructive conversations,

83
00:03:37,760 --> 00:03:39,120
Speaker 2: and we would like to hear what they have to

84
00:03:39,120 --> 00:03:41,360
Speaker 2: say as part of the public inquiry before we make

85
00:03:41,400 --> 00:03:42,400
Speaker 2: any further decisions.

86
00:03:42,520 --> 00:03:44,320
Speaker 1: Okay, what are you going to do about this brown

87
00:03:44,360 --> 00:03:45,120
Speaker 1: onion business?

88
00:03:46,720 --> 00:03:49,840
Speaker 2: Yes, so we are aware of that situation. I mean,

89
00:03:49,960 --> 00:03:54,640
Speaker 2: ultimately it is it is theft, but look, we trust

90
00:03:54,640 --> 00:03:56,200
Speaker 2: our customers to do the right thing.

91
00:03:57,880 --> 00:03:59,920
Speaker 1: Noticed the evidence would suggest that you shouldn't.

92
00:04:01,240 --> 00:04:05,080
Speaker 2: Look, we haven't noticed any unusual increase in brown onions

93
00:04:05,120 --> 00:04:07,960
Speaker 2: going through self checkouts. We trust our customers to be

94
00:04:07,960 --> 00:04:10,400
Speaker 2: honest and most people do understand that theft is theft

95
00:04:10,440 --> 00:04:11,320
Speaker 2: and we'll do the right thing.

96
00:04:11,680 --> 00:04:14,040
Speaker 1: Yeah, all right, Julian listen, thanks very much, really appreciate

97
00:04:14,040 --> 00:04:16,400
Speaker 1: your time. That's Julian Benefield, the General Council for Food

98
00:04:16,400 --> 00:04:20,239
Speaker 1: Stuffs North Island. For more from Heather Duplessy, Alan Drive,

99
00:04:20,400 --> 00:04:23,800
Speaker 1: listen live to news Talks. It'd be from four pm weekdays,

100
00:04:23,920 --> 00:04:26,120
Speaker 1: or follow the podcast on iHeartRadio